A/B Street: A simulation game to fix Seattle's traffic • Hacker News Board

Re: A/B Street: A simulation game to fix Seattle's traffic

#21

Post by com2kid » Mon, Jun 22, 2020, 10:12 PM UTC

sytse wrote: post #12

If you like this game consider checking out Cities Skylines. Traffic is a major part of the game. With the vanilla edition you can already do a lot, but if you want to route lanes you need the traffic manager mod https://steamcommunity.com/sharedfiles/filedetails/?id=16376...

Do any of the mods give the simulated people real commutes? I gave up after I made a city with a full subway system and zoned areas and watched as the simulated citizens left home, went to work, and then went to another random home at night.

It completely ruined my fancy subway system's timetable. :(

Re: A/B Street: A simulation game to fix Seattle's traffic

#22

Post by petermcneeley » Mon, Jun 22, 2020, 10:12 PM UTC

dabreegster wrote: post #2

Hi, author here. Previously posted to HN ( https://news.ycombinator.com/item?id=21763636 ). I just released alpha -- check out this cheesy trailer ( https://www.youtube.com/watch?v=LxPD4n_1-LU ). Happy to answer questions! And if you're serious about running this in your city, the importing process has improved greatly since December, so get in touch.

I wonder what the how far the optimal is from existing for variations of global vs greedy (local) optimization.

Re: A/B Street: A simulation game to fix Seattle's traffic

#23

Post by flr03 » Mon, Jun 22, 2020, 10:14 PM UTC

dabreegster wrote: post #2

Hi, author here. Previously posted to HN ( https://news.ycombinator.com/item?id=21763636 ). I just released alpha -- check out this cheesy trailer ( https://www.youtube.com/watch?v=LxPD4n_1-LU ). Happy to answer questions! And if you're serious about running this in your city, the importing process has improved greatly since December, so get in touch.

Awesome job congratulations this looks it detailled, I can see there is a lot of work behind.

Re: A/B Street: A simulation game to fix Seattle's traffic

#24

Post by StillBored » Mon, Jun 22, 2020, 10:19 PM UTC

This is completely awesome. I've considered doing something similar (mostly while sitting around in Austin looking at traffic lights backing up 100+ cars to allow 2-3 to enter from a side street) or looking at a pair of lights (accidentally?) acting like a metering device. Particularly city wide as I'm 100% convinced that the actual traffic engineers here are doing something wrong. They reworked a bike lane a few years back on a street I was driving on daily and then I could see them out there for months trying to work around the fact that they backed up a couple major intersections a couple blocks away as a result.

So the question is, given a bit of ML/etc driving it, and some actual commute time data, what are the changes something like this actually can be used to improve traffic?

Re: A/B Street: A simulation game to fix Seattle's traffic

#25

Post by MasterScrat » Mon, Jun 22, 2020, 10:24 PM UTC

If you're interested in this kind of planning problems, check out the NeurIPS Flatland challenge, in which we try to optimize traffic on railway networks, in association with the Swiss and German railway companies!

https://www.aicrowd.com/challenges/neurips-2020-flatland-cha...

(disclaimer: I'm an organizer)

Re: A/B Street: A simulation game to fix Seattle's traffic

#26

Post by linuxftw » Mon, Jun 22, 2020, 10:32 PM UTC

The problem is there's only so much room for automobiles in a city. The more you decrease traffic, the more you'll induce demand to fill those improved traffic flow patterns.

The more infrastructure that is built downtown, the more people want to be downtown.

Re: A/B Street: A simulation game to fix Seattle's traffic

#27

Post by andrewxdiamond » Mon, Jun 22, 2020, 10:32 PM UTC

yazaddaruvala wrote: post #5

I've been really hoping for the Maps team at Google/Waze to release a similar "traffic congestion" view to highlight consistent hot spots. A simulation is also pretty good, but the real data from Maps/Waze users will be even better! Ideally, a government dashboard to see traffic congestion changes per year, month, day of week, but also hour. If we can create metrics for cities like p99 door-door trip time (i.e. laten…

Your hope has come true! Kindof. I don't think this comes with a dashboard, but the data is there

https://www.digitaltrends.com/cars/waze-for-cities-data-puts...

Re: A/B Street: A simulation game to fix Seattle's traffic

#28

Post by dabreegster » Mon, Jun 22, 2020, 10:34 PM UTC

StillBored wrote: post #24

This is completely awesome. I've considered doing something similar (mostly while sitting around in Austin looking at traffic lights backing up 100+ cars to allow 2-3 to enter from a side street) or looking at a pair of lights (accidentally?) acting like a metering device. Particularly city wide as I'm 100% convinced that the actual traffic engineers here are doing something wrong. They reworked a bike lane a few yea…

There are lots of data quality and simulation assumption issues that prevent results from being meaningful. I'm mostly positioning this as a way to get people interested in coming up with specific ideas, but of course I'd love to reach trustworthiness parity with industry standard traffic sims.

I actually got started in traffic sim while I was in Austin! If you know anybody there who'd be interested in putting in some work to get the area running smoothly, I'd love to include it as a second city in the game.

Re: A/B Street: A simulation game to fix Seattle's traffic

#29

Post by kevmo314 » Mon, Jun 22, 2020, 10:37 PM UTC

dabreegster wrote: post #18

evolve2k wrote: post #8

How much would be involved to adapt this to other cities?

https://github.com/dabreegster/abstreet/blob/master/docs/new... Biggest barrier is lack of trip demand data. If you come up with how many people leave from building1 to building2 at time T using some mode of transport, then this can work.

I see you worked at Google at the same time I was there. I'm surprised we never got in touch, I helped work on a project that did traffic light optimization research, also investigating some of the approaches you mentioned, such as microbidding. We didn't build our own microsimulator but instead worked off of SUMO + OSM, which I'm sure you're familiar with. :)

I no longer work there either, but feel free to reach out if you're interested in chatting.

Re: A/B Street: A simulation game to fix Seattle's traffic

#30

Post by asdff » Mon, Jun 22, 2020, 10:37 PM UTC

cyral wrote: post #16

sytse wrote: post #12

If you like this game consider checking out Cities Skylines. Traffic is a major part of the game. With the vanilla edition you can already do a lot, but if you want to route lanes you need the traffic manager mod https://steamcommunity.com/sharedfiles/filedetails/?id=16376...

Cities Skylines with the traffic manager mod is a great way to spend a whole lot of time if you are bored. It’s surprisingly fun to try to optimize traffic.

And surprisingly frustrating. Why is my city gridlocked? Because only the left most lane of my 6 lane avenue is being used by these 1000 hot dog trucks leaving my port.