This is super cool, thank you for building it! Two small UX ideas:
- a scrollbar and search for the Online Library would be helpful
- switching difficulty levels in the middle of reading could be helpful. Or if you keep that on a separate page, returning automatically to the last open position. (I was floating between beginner levels to find the right amount of challenge)
Super awesome! I like how you just color roads to show time. When you calculate polygons to try and cover the whole area in some 5-10 minute bucket, you can wind up with all sorts of odd holes far away from roads. Keep it simple.
Hi! https://a-b-street.github.io/docs/project/history/retrospect... has a summary of where the project went. The original scope of the traffic simulation was too grandiose, and it was hard to influence real decision-making with something so complex and buggy. A/B Street morphed into several simpler tools focused on specific problems (removing through-traffic through an area, estimating uptake of new cycle lanes). I'm focused more on web-based tools for gov use these days, but will get back to simulation someday!
In my experience, the published paper is super vague on the approach, and implementing it without further references is really hard. I'm not necessarily arguing that papers should get longer and more detailed to counter this; expressing the details that matter in code seems like a more natural way to communicate anyway.
Why trust results if you can't see the methodology in detail and apply the approach to your own data? I once knew somebody who built a fuzz tester for a compilers project, got ahold of a previous project's compiler code that won best paper awards, and discovered a bunch of implementation bugs that completely invalidated the results.
Why is the peer review process limited to a handful of people who probably don't have access to the code and data? If your work is on Github, anybody can come along and peer review it, probably in much more detail. And as a researcher, you don't get just one chance to respond to their feedback -- you can actually start a dialogue, which other people are free to join in.
As long as a project's README makes any sort of quality / maintenance expectations clear upfront, why not publish your code?
abstreet.org has an offline import process that combines data from OpenStreetMap, city-specific GIS datasets, and elevation into a single file. The process has to be deterministic, given the same input and code, and calling out to an external API is a non-starter. We use https://github.com/eldang/elevation_lookups, a Python library that downloads missing LIDAR or SRTM data and uses GDAL to handle batch queries. Two issues with it are having too many dependencies (so we run it in Docker) and not being able to parallelize lookups without blowing up memory, due to some GDAL caching internals.
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.
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.
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.