Sumo – Simulation of Urban Mobility
eclipse.dev
eclipse.dev
Example:
For over a decade, the freeway on-ramp nearest my work had two main ways of getting to it from downtown. One of them involved a stop-sign crossing a road that had the right-of-way (i.e. a two-way stop). The other had timed traffic signals. Every evening around 5pm, the traffic would backup from the stop-sign for multiple blocks. Meanwhile the route with lights was completely smooth.
Eventually the stop-sign was replaced with a signal, but I marveled at how many people persisted in making their daily commute much worse than it needed to be.
Other groups of people are incredibly pessimistic and always take the least bad route.
For example, people often have a preference for certain flow patterns in traffic that they prioritize more than the route with the shortest time.
To this day, people still slow down like they used to as if it were a 55.
Imaging the simulation being running headless, decoupled from the GUI client
If you want traffic that moves in real time, then your day/night cycle needs to also be real time: 24 hours. That means that if you're only playing in 2 hour gaming sessions, it's going to take nearly 2 weeks to simulate a single day. For those that truly want realism, that's great. For most gamers though, that's gonna be a non-starter.
So of course there's a problem: You can either have traffic moving in real time, or you can have a reasonable day/night cycle length. You can't have both.
The compromise that city builders make ends up with what you said: Ambulances take 3 months to reach patients, even if they're only traveling a mile.
That said, I do think that Cities: Skylines could do it better. The amount of semi-truck traffic in that game is absolutely insane. Easily 4x what it needs to be. You make an industrial district that's only ~1/2 of a square mile and it's an absolute zoo of semi-trucks, requiring some crazy traffic engineering to make it not be a gridlock.
People are going to sweep parameters on the cluster.
"Release date: Jan 26, 2021"
actually looks really fun tho
The Projects page is worth looking at too.
I’ll have to give this a test run later.
Anything serious needs to include those - what would Amsterdam look like if all those people stopped riding their bikes and got in cars? It'd be a disaster.
The state/city engineers only cared about efficiency of car traffic. bike and ped infrastructure is not included or meet bare minimum spec (ie, "share-ow" lanes/paint/signs on busy streets as a very poor substitute for bike infrastructure)
Google Maps (or others) works good to find a resonable route, but I can do better on my own. One-way streets where bikes are allowed to go do opposite way is sometimes missing, short desire paths connecting bike ways, crossings where it's safe to do an (illegal) right-on-red etc.
Tried a GDPR data claim from Voi but got nothing back :( But I hope the data is somehow available for urban planners, think it would be a great source of truth to use in tools like this.
The most important input to a traffic simulator such as this is the so-called "traffic demand", i.e. the routes that vehicles follow. Typically this is provided in the form of origin-destination matrices, but this data is not freely available.
Next up is the way in which traffic lights work. Reality is very hard to model here, again because the data is not freely available.
And then, due to numerous modeling errors in vehicle density, in the way that roads differ from e.g. OpenStreetMap, and how traffic behaves, the simulations are highly unrealistic, unless one spends some time to calibrate it.
It costs quite a lot of money to set up a realistic simulation, and most governments use commercial tooling that is easier to use, such as Vissim or Aimsun.