This isn't a system to detect if there are cars at an intersection. Wire loops and other systems already can and do measure cars like that.

This is a system to detect "in 32 minutes, twenty-five cars will come from the south, crossing sixteen cars from the east. By pausing the western traffic lights for three seconds, both streams of cars can drive past each other without clogging up the main intersection".

Users don't need to press buttons on apps, they're just using their existing navigation apps which are already providing them with directions (or in the case of Flitsmeister, speed cameras). Similar apps also exist for pedestrians and bikes in my country. My biking experience certainly has improved by having lights on empty/near empty intersections automatically turn green when I'm approaching.

I'd rather see more roundabouts rather than the badly scheduled traffic light system my city has, but the system works for me.

There's the ever-present privacy threat of sharing your (aggregate) location, of course, but in this case we're talking about the government, which pretty much has a live map of what phones are moving where already.

Is that level of complexity actually necessary? The way that the UK system appears to work (as a user) is that the IR sensor detects a vehicle approaching the intersection from the minor road and, if there are no vehicles approaching on the major road, it gives you a green light by the time you actually reach the junction.

I agree that the app-based system would theoretically be slightly better in that it has more information to work with, but given that we're basically talking about a stochastic process then it feels like the IR system should really be good enough.

It's not strictly necessary but it takes giving a crap and effort and constant auditing and reassessing to do all that light timing well with the current state of technology. Outsourcing that responsibility to centralized computer that can do most of it automatically and on a schedule is likely a huge improvement for the quality of programming/timing for the average light if only a neutral for the best ones.

Predict the exact traffic situation 32 minutes in the future? Where would the information come from? The actual system in Flanders predicts about 1 minute, and that only for the main arteries. And the system barely uses the info from the vehicles, the main source is induction loops. Only a small percentage of the vehicles send their location, and hardly any cyclist and pedestrians. Vehicles don't send their destination, only the 'turn-intention' on the next intersection. But even that is unreliable, so typically not used.

Do you know if there is any documentation on how they actually implement the algorithms behind it?

Seems like possibly it's a small enough region that you could cram it into a MILP solver?

Thank you, this is a much better explanation than in the article.

That's not how it works - the idea is your navigation app signals the lights in advance. If you will reach red lights in 1km, the app signals this and the lights will be green before you're there, so you don't need to slow down.

Most people do not use a navigation app for most trips (i.e. commuting to work, grocery shopping, or whatnot).

Cyclist aren't massive nor hot, on a hot summerday they will be cooler than the road surface. I think a camera with object recognition or something similar should be able to do it.

It's not the first time apps are being shoved where they shouldn't belong by the Flemish Government. They made a cooking app of half a million Euro [1] and there were trials where you could use an app to to get the deposit back on plastic bottles by scanning the bottle and your wastebag [2].

I'm not sure if it's incompetence, trying to look modern or just plain old corruption.

[1] https://www.vrt.be/vrtnws/nl/2025/05/27/kookapp-van-vlaamse-...

[2] https://www.vrt.be/vrtnws/nl/2023/05/23/cordacampus-proefpro...

> Cyclist aren't massive nor hot, on a hot summerday they will be cooler than the road surface. I think a camera with object recognition or something similar should be able to do it.

I don't know the technical details on how it's done here in Stockholm, even less in the suburbs where I've lived but it feels like magic that in some traffic lights in busier intersections they'll just trigger by themselves with enough distance that when I arrive with the bike at the intersection the lights are turning green for me.

I have looked for cameras but couldn't find them, it's even more puzzling in a specific intersection close to my house where I bike along a 4-lane road with only fields of wheat and grass, and the intersection I cross leads to an industrial area, there are only the posts for the pedestrian/bike lights and those still turn green, no cameras in sight, no detection trap on the pavement.

In Stockholm, we have 2 sets of sensors dug :

1. induction coil sensor dug in road at the traffic light

2. radar sensor at 100-200m from the traffic light

It feels like magic because if you hit sensor (2) and there are no other competing cars, the traffic light changes for you and you don't have to slow down.

This is the best solution. In Dublin, they only have sensor (1) and it feels really slow in comparison. You keep having to stop and start.

Indeed, here in Linz the few intelligent traffic lights we have are mostly for cyclists and the detection works great. On the three bridges there are counters that differentiate between trucks, cars, cyclists and pedestrians and they also work great.

I actually assumed that this is because the system tries to figure out where you will be by your planned route. so seeing the bike coming up is much later - for example when the bike will be coming around a corner

> Having drivers reach for their phone when they're approaching traffic lights - common pedestrian crossing points - is categorically moronic.

The article says that the they just ned a relevant app running on their phone when they're approaching an relevant traffic light, and there's no mention that users need to actively do anything.

Also please have a read through the HN guidelines regarding your punchline "is categorically moronic."

> When disagreeing, please reply to the argument instead of calling names. "That is idiotic; 1 + 1 is 2, not 3" can be shortened to "1 + 1 is 2, not 3."

https://news.ycombinator.com/newsguidelines.html

How is this categorically wrong comment at the top of the thread?

Many cities use TPMS monitoring/fingerprinting on highways to gauge traffic levels. Wouldn't be shocked if they started using that for traffic light timings

Do bicyclists and pedestrians really require the use of sensors? There are a lot of paths around me that cross busy roads with traffic signals. We just press a button. It speaks after you press it for visually impaired users and shows a red light after you press it for audibly impaired users.

Do pedestrians need automatic detection? Not necessarily, although a beg button adds unnecessary friction.

Do cyclists need automatic detection? Yes, absolutely. If you think otherwise, would you want to park and exit your car every time you want to cross an intersection? It's no different for a cyclist.