2 ms·

This is real great engineering! Tldr key facts: - be smart and detect actual traffic for modifying the red light behavior in real time - for example, switch

This is real great engineering! Tldr key facts: - be smart and detect actual traffic for modifying the red light behavior in real time - for example, switch to green immediately for walkers if there is no car even if the displayed countdown is not yet finished. - operate independently each red light of a section to not need to have them operate stupidly in symmetric/synchronized state machine cycles. Don't just give a priority to car or bicycles, but try to ensure that everyone get the less friction and maximum mobility possible.

simion314 6y ago
I think you should also add a point for public transport, a bus with 30 people should not have the same priority like a car with 1 person in it. Trying to keep public transport always moving will also have other benefits.
- mrandish 6y ago
  That's an interesting idea, however it could lead to some unintended consequences such as cars being incentivized to 'caravan' with busses.
  - notjustbikes 6y ago
    Transit should always be in it own lane (as it is in most of the Netherlands) so there's no downside to this.
- wjnc 6y ago
  Solved as well. Special bus lanes exist and have separate lights that (seem to) get priority in these smart systems, at least over cars. The fun thing is that these are often routed parallel to the bike routes, at least in the few examples I can think of right now, so if the bus comes and gets priority the cyclists get green as well. Never thought about that our traffic lights could be something special.
mrandish 6y ago
> "be smart and detect actual traffic for modifying the red light behavior in real time" This is something I've been frustrated by for years. "Autonomous Cars" are hard. Autonomous traffic lights should be much easier. Here in the suburbs there are tons of stoplight intersections between small residential neighborhoods and central arteries where cross traffic is fairly rare. There are sensors in the roadway to detect cars waiting but they are still incredibly slow to change as they seem to stick to minimum timers during most of the day. The crazy part is that one or two cars waiting to cross often wait two minutes for the apparent timer to expire and the light to change while there was no oncoming traffic the other way that would have been inconvenienced by an immediate change. The cross-traffic cycles are very short anyway because there's only a couple cars at most. Even more perverse is the number of times I sit waiting two minutes with no approaching cross traffic in sight from either direction, then when some cars are finally approaching, then it stops them with a red so I can cross. A semi-smart system based on a Raspberry Pi-level SBC with a camera would have had me on my way while no one was coming. It seems like in the aggregate camera-based autonomous traffic lights could save enormous amounts of fuel and time.

simion314 6y ago

I think you should also add a point for public transport, a bus with 30 people should not have the same priority like a car with 1 person in it. Trying to keep public transport always moving will also have other benefits.

mrandish 6y ago
That's an interesting idea, however it could lead to some unintended consequences such as cars being incentivized to 'caravan' with busses.
- notjustbikes 6y ago
  Transit should always be in it own lane (as it is in most of the Netherlands) so there's no downside to this.
wjnc 6y ago
Solved as well. Special bus lanes exist and have separate lights that (seem to) get priority in these smart systems, at least over cars. The fun thing is that these are often routed parallel to the bike routes, at least in the few examples I can think of right now, so if the bus comes and gets priority the cyclists get green as well. Never thought about that our traffic lights could be something special.

mrandish 6y ago

> "be smart and detect actual traffic for modifying the red light behavior in real time" This is something I've been frustrated by for years. "Autonomous Cars" are hard. Autonomous traffic lights should be much easier. Here in the suburbs there are tons of stoplight intersections between small residential neighborhoods and central arteries where cross traffic is fairly rare. There are sensors in the roadway to detect cars waiting but they are still incredibly slow to change as they seem to stick to minimum timers during most of the day. The crazy part is that one or two cars waiting to cross often wait two minutes for the apparent timer to expire and the light to change while there was no oncoming traffic the other way that would have been inconvenienced by an immediate change. The cross-traffic cycles are very short anyway because there's only a couple cars at most. Even more perverse is the number of times I sit waiting two minutes with no approaching cross traffic in sight from either direction, then when some cars are finally approaching, then it stops them with a red so I can cross. A semi-smart system based on a Raspberry Pi-level SBC with a camera would have had me on my way while no one was coming. It seems like in the aggregate camera-based autonomous traffic lights could save enormous amounts of fuel and time.