Transforming OpenStreetMap into thick roads and intersections
a-b-street.github.ioI am a professional traffic engineer and I build micro-simulation models for most of the projects I am involved with. I can say that compared to software that is used in industry (PTV Vissim, PTV Vistro, and Trafficware Synchro) it looks like A/B is a reasonable toy model that can get pretty close to the real thing. I think it is great that software exists at a level for individuals without huge budgets to be able to build and play around with traffic networks. Automating the intersection setup goes a long way toward this type of tool being accessible to the lay person. All that being said, I think the limitations of any model are important to understand when interpreting the outputs. I can take any model and make traffic flow or decrease the delay per vehicle, that doesn't mean my results are realistic.
Keep up the work, this is an awesome tool and I hope it can get to the point where it can easily help inform people about traffic design and simulation.
Thank you for the industry insight! I'm definitely not trying to match the degree of realism and calibration with the professional stuff -- there's no way I could, without way more resources. A/B Street's simulation is more for visualization than getting meaningful results -- although I am quite proud of the dataviz we've baked in.
From talking to planners, reaching for these tools is a huge time and money commitment, even assuming they have a license and somebody trained to use it. Some ideas don't even get off the ground, and some projects take community feedback after spending months building the initial models. I think there's a huge missing space for rapid prototyping. The fact that I see planning agencies regularly include graphics from streetmix.net is evidence of this -- it's a quick way to communicate, so often how a conversation starts. I'd simply like to expand that space.
At the end of the day better communication between the public, engineers, planners, and respective agencies is the goal. We have a joke that gets repeated all too often, everyone is a traffic engineer, just ask them! Rapid iteration is key to be able to look at all scenarios, no matter how crazy they may be.
What kinds of feedback does your profession care about? I am constantly frustrated as a pedestrian with the traffic flow and signal timing problems in my area but I feel like it's pointless to submit a complaint to my local municipality because my anecdotal data points will probably be disregarded.
I'm not very good at nor patient for trying to do neighborhood people organization and go around in circles with City Hall for 5 years (as has been the case with other neighborhood things like park use etc. that I have previously been involved in). But I am tried of losing about 5-6 hours a year to excess wait times at intersections despite trying to avoid the slowest crossings.
I've lived a few different places, also working in transportation engineering. I personally care a great deal about accessibility, and would much rather delay cars slightly than pedestrians (for numerous reasons). For example, when reviewing traffic control plans, etc, I always make comments or suggest changes along those lines. But ultimately, it comes down to what the municipality wants to prioritize - unless there are safety issues, it's hard to do more than suggest and push for something better.
If you want to make a difference, making comments (about ongoing projects, particularly) will tend to have more impact. For something more systemic (like light cycles deprioritizing pedestrians), I found the best solution was to move to a better city.
If that's not a solution for you :) then another option you might want to consider is having a meeting with your municipal (or even higher) level elected representative - invite them for a walkthrough of a more problematic space, and try to explain the issues and possible solutions as calmly/rationally as possible. And without attacking car culture directly, as that tends to get people defensive.
Thanks a lot for the feedback. I live near a traffic artery so I've always known there were trade-offs, but the frustration does mount after a while. My #1 beef is when the traffic is paused but you just missed the light change when you hit the ped xing button, and now it's not safe to rely on the light staying green while you dart across.
San Francisco avoids this particular problem by having the walk signal turn on automatically when the car signal is green — no need to hit a button (most intersections don't have buttons at all, and the ones that exist aren't even hooked up to the light).
I don't know if this would translate well to other cities — SF has a very pedestrian-friendly layout with narrow, slow streets , small blocks, and short light cycles — but it works well enough here that it might be worth trying elsewhere.
When I visited Barcelona, I was struck by their intersection design. It's great.
Instead of squares, their streets intersect in diamonds, where the (largely one-way) streets come out of the corners. They're a bit larger than a square intersection-- it's like you took the square, and then filled in four right triangles at each of the corners.
The main effect is that crosswalks are significantly pushed back from the intersection center. As a pedestrian, there's far less ambiguity about where the cars are going (regardless of whether they use their signal). Same for drivers: you never have to wonder which way a pedestrian is gong to cross when they're standing at a corner where two crosswalks meet, because there are no such corners.
Of course, this also makes intersections much bigger. This is offset by having parking spaces along the edges of intersections.
That might be great if you're visiting or just out for a stroll.
But let me tell you, it can be infuriating if you're walking to actually get somewhere. The city I live in has been replacing 2-way and 4-way stops with roundabouts, which has a similar effect of moving the crosswalks away from the intersection. Now it takes much longer to walk anywhere because you keep having to detour out of your way to cross the street.
Yes, 'beg buttons' are awful (as are the 'red-revert' signal designs you allude to).
Can you or any other traffic engineers here tell me why we find it acceptable that cars are allowed to make unprotected left turns while pedestrians have a walk signal? Right turn on red is bad enough, but unprotected lefts give drivers a direct incentive to torpedo into crossing cyclists / pedestrians, as slowing down would mean a collision with oncoming traffic! Too many intersections in downtown Atlanta, for instance, have no left turn arrow, and I felt constantly in danger.
Most North American road infrastructure has been rebuilt over the last century to be automobile-centric. "We" deem that conflict to be acceptable because the alternative is more complicated phasing - most of which results in delays to car traffic (and pedestrians!), and so most places won't accept it. For details, here's [1] a document that outlines the tradeoffs in various different signal phasing designs.
TL;DR is that "we" won't prioritize any road user except the automobile.
[1] https://ops.fhwa.dot.gov/publications/fhwahop08024/chapter4....
I agree completely with this comment. Most places signal timing is done with the main scoring metric of reducing vehicle delay. Pedestrians are just an after thought that can be squeezed into a vehicle specific timing plan. As an example, how often do you see a pedestrian scramble, walk signals in all directions, outside of somewhere like Time Square?
Some of the problematic intersections in Atlanta have slowly been replaced with scrambles, and it's amazing as a pedestrian. I hope to see more and more of those!
As someone with interest in traffic models, what spaces do you or your peers hang out outline?
That is a tough question because I can't think of any places really. Maybe look into the Institute of Traffic Engineers, they may have a forum of some sort.
A/B Street is an open-source traffic simulator that lets you edit roads and intersections, based on OpenStreetMap data. But to even do that, first we have to geometrically represent the transportation network in great detail. This is a second deep-dive into how things work.
And in case you're interested how the traffic simulation part works in detail, but missed yesterday's post: https://news.ycombinator.com/item?id=28553654
You can play it right in your browser: http://play.abstreet.org/0.2.58/abstreet.html What a cool project!
Years back as OSM was fading and the v1.0 of mapbox started rolling out my pedestrian view of the space was that we were seeing the endgame unfold
It's insane to see the levels of depth still to plunge - this is incredible documentation and a very cool project
PS - I'd upvote twice if I could for incorporate of cthulhu mythos!
Just this morning I came across new work from the Berlin OSM community: https://supaplexosm.github.io/strassenraumkarte-neukoelln/?m... The end-game isn't in sight. We (data consumers of OSM) are just getting started.
Supaplex did supplement their map with open data from the municipality though for the street outlines as far as I understand (I've asked), so it's not 100% OpenStreetMap data used there.
Totally awesome and inspiring nonetheless though.
Right, from https://supaplexosm.github.io/strassenraumkarte-neukoelln/po... (translation https://translate.google.com/translate?sl=auto&tl=en&u=https...), there was an extra curb data-set used. I've actually found a few of these (https://a-b-street.github.io/docs/tech/map/geometry/index.ht...), so the approach is maybe repeatable.
Regardless, the design choices and color scheme are something I would love to adapt.
You can definitely map curbs in OSM if you have high resolution imagery: https://wiki.openstreetmap.org/wiki/Key:area:highway
Do not expect however mainstream data consumers to support it ;)
That is gorgeous! I want it.
What to you mean, "OSM was fading"? When?
I'm also curious about this. If you look at the number of users, number of edits, etc then openstreetmap looks like it is on a rocket trajectory.
More and more corporations are putting significant amounts of energy towards improving openstreetmap as well, drastically improving overall quality.
It's scary to think such a great project could have perished at some point in its infancy. It's not perfect yet but it's an incredible resource and alternative to GMaps.
> Highway on/off ramps https://a-b-street.github.io/docs/tech/map/geometry/index.ht...
It's hard to say which is better here, to be honest. A giant blob of gray is less helpful than showing some lanes, but the original version is much closer to the actual shape: https://www.google.com/maps/@47.6564275,-122.3223112,203m/da...
Though apparently even google can't handle this intersection. Where did the middle layer of road go, google? https://www.google.com/maps/@47.6556467,-122.3191587,69a,35y...
Hmm, I guess I never checked satellite for that example. Part of the motivation there was improving the simulation. While vehicles are moving through the intersection, they grab a lock and prevent conflicting movements. A huge intersection takes longer to cross and was causing unrealistic backups here. But that was only because of the wacky lane-changing model (can only change between those 4 northbound lanes in an intersection, where conflicts are modeled) and poor lane placement at the time. Some of that's been fixed, worth revisiting the shape.
Very cool. Unfortunately it looks like I managed to put it into an infinite loop in the tutorial. Or possibly I just didn’t wait long enough after saving my changes…
I haven't given any love to the poor tutorial mode in ages, so I'm not surprised you found a bug. Please file an issue (https://github.com/dabreegster/abstreet/issues) and describe how to reproduce the problem. If you can open the browser developer console (Ctrl+shift+I in firefox), often the last part will be really helpful.
This is the coolest project I've seen in a long time. I cannot wait to see how this evolves and scales up.