An end-to-end look at mapping pedestrian infrastructure in OpenStreetMap
This is a modal window.
The media could not be loaded, either because the server or network failed or because the format is not supported.
Formal Metadata
| Title |
| |
| Title of Series | ||
| Number of Parts | 41 | |
| Author | ||
| License | CC Attribution 3.0 Unported: You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor. | |
| Identifiers | 10.5446/58236 (DOI) | |
| Publisher | ||
| Release Date | ||
| Language |
Content Metadata
| Subject Area | ||
| Genre | ||
| Abstract |
|
State of the Map US 20222 / 41
1
4
7
8
14
18
21
22
24
27
28
30
37
39
00:00
World Wide Web ConsortiumOpen setRoutingMappingSurfaceAdditionAndroid (robot)Covering spaceMobile WebNetwork topologyQuicksortGraph (mathematics)Service (economics)BuildingSoftware developerPixelMedical imagingPoint (geometry)SoftwarePolygonNear-ringAreaLevel (video gaming)FrustrationInheritance (object-oriented programming)Coefficient of determinationUniform resource locatorBlogPlanningInformation technology consultingDefault (computer science)BitSatelliteView (database)FacebookElectronic visual displayTerm (mathematics)InformationNatural numberDigital photographyMotion captureField (computer science)State of matterCurvatureLocal ringPopulation densityProbability density functionSpring (hydrology)Data storage deviceDifferent (Kate Ryan album)PredictabilityText editorBlock (periodic table)Right angleMetreTwitterSelf-organizationWavePrisoner's dilemmaGroup actionCuboidAddress spaceType theory2 (number)TrailKey (cryptography)Software frameworkComplete metric spaceCentralizer and normalizerTransportation theory (mathematics)Characteristic polynomialMereologyMeta elementGoodness of fitMaxima and minimaComputer animation
Transcript: English(auto-generated)
00:04
All right, well, good morning, everyone. It is a pleasure to be back at a State of the Map conference. This is my first State of the Map US. Thanks for having me, and thanks to the organizers for putting it on. We heard this morning how difficult it was with everything that's happened over the last few years. My name's Eduardo.
00:21
I am here representing META, talking about an end-to-end look at pedestrian infrastructure in OpenStreetMap. I've been involved with OpenStreetMap for about seven years now. It's just a pleasure to be working in this community. So I want to talk today about pedestrian infrastructure.
00:40
It's something that I'm pretty passionate about. I spoke about it last year at the online conferences of Phosphogee and State of the Map. So if you go back to 2021, I was talking about mapping pedestrian infrastructure, but particularly looking at five different cities. I looked at Folsom in the United States. I looked at Heidelberg in Germany,
01:00
Melbourne in Australia. I looked at Stonetown in Tanzania and Yesan in South Korea. These are cities with very different characteristics, different populations, different levels of economic development, different cultural backgrounds. But all of them basically had the same thing where the data in OpenStreetMap did not necessarily mirror what was on the ground.
01:22
So we can do a lot better to map pedestrian infrastructure, but also that the infrastructure itself can be improved. And I think that's true in most cities around the world. Pedestrian infrastructure is neglected. So yeah, here is Folsom. This is where I've been living since January last year
01:42
when I moved to the United States. So I was actually living here in this particular spot for about a month when I moved to the United States. And even just going out to get groceries, like walking 200 meters to get groceries was a bit of a pain. Like it was all designed for cars, even if the grocery store was just across the road. And that was very interesting for me.
02:02
And I started thinking a lot more about pedestrian infrastructure and how the tools that I work with daily and that a lot of you guys work with daily can be used to make pedestrian infrastructure better. So there's a lot of tags related to pedestrian data. Like the obvious stuff is listed here. I don't know how well you can see it,
02:20
but we're talking things like sidewalks, we're talking things like crosswalks. This is near Folsom Prison, which you might know from the Johnny Cash Show and the dam itself. Really nice suburbia, got really good sidewalks that kind of are perfect for recreation, but not so much for getting down to your local shops.
02:40
Building entrances, another one related to pedestrians. If you're going A to B, rather than just routing to the centoid of that polygon or trying to work out which edge of the building is relevant, actually knowing where the entrance is is really good for large buildings. But there's more than just that. We need to think more broadly when it comes to pedestrian data.
03:01
How many people have seen this you know, running or walking on the local trail? This is, I don't know how well you can see it. It's probably what you think it is. You know, when your four-legged friend does this in the middle of the track, it can be very annoying. And that's why it's great. The city of Folsom also has,
03:21
they've provided bags for dog excrement. So OpenStreetMap is a great community. We have tags for this. You can map it. This is another example of something that pedestrians care about that we can have on the map. And so you can know where this is. You can map the trash bag location
03:42
so that you can get a trash bag when your dog does this. You can pick it up and then just leave it there. Super frustrating. So in all seriousness, I wanted to map this particular area, the place where I first moved to in the US,
04:01
just a temporary accommodation. It's a shopping area in the south part of Folsom and near Highway 50 for those that have driven from San Francisco to Tahoe. And there was really no pedestrian data in OpenStreetMap. It was mapped really well as far as like the road network but again, no sidewalks, no crosswalks, nothing like that.
04:22
So I started mapping that. There's a closer look at what was there before I started mapping. Routing actually worked really well. So credit to the open route service from Heidelberg University actually just tries to work out using the underlying road data, what would it be like for a pedestrian.
04:41
It seemed to work fairly well. I guess it's not always true for longer routes but this worked well. Graph Hopper worked well. So credit to both of those routing engines. But I wasn't interested in the routing. I'm interested in the actual pedestrian infrastructure. And so I started with Bing imagery, which is great. So thank you Bing, thank you Microsoft. That's great imagery to work with.
05:01
You could actually see, if you look at this image, you can see the crosswalks very clearly. You can see the sidewalks clearly. You can even see like tactile paving in some scenarios. So using ID or rapid, you can map that out fairly well. Sidewalks were starting to appear once I was mapping this, which was great.
05:22
But the problem is that you have some tree cover. You have building awnings that block out a lot of the views that you would see from a satellite or aerial image. And so you had to do more field work to get the data I needed. So that's where I looked at Mapillary. Mapillary is a tool for street level imagery. I think Antonia mentioned it before.
05:40
So thank you for that. The coverage there wasn't great initially. So I decided to go out with a 360 camera and capture more. This is the GoPro Max, which takes a photo every two seconds, which is really good when you're going at like a slower speed. It's actually good enough for this purpose. Ironically, I was driving through the area
06:01
to map pedestrian infrastructure, but that's the nature of this place. Really good, like in terms of the setup, the front lens worked really well here. The back lens, like I didn't have it mounted high enough on my vehicle. So the back lens wasn't so great, but still incredibly useful. I had the iPhone. Oh, that's just a look at the full view.
06:22
And I had the iPhone mounted in the front to capture like high detail. This is good for like building, I guess like the name of the building and addresses, things like that that might be useful for pedestrian routing. Most 360 imagery, you can see like very clearly here, even on the right where there's a building awning that blocks things from satellite image,
06:41
you can start to map that out. With Mapr, you can go direct to Rapid or ID Editor to actually start editing this. So here we're looking at Rapid. You can see there's not too much like ML predictions here, most of it's been added already, but you've got the street level imagery, which is great for the building entrance.
07:00
So we can add that here on the building polygon. And the Mapillary detects crosswalks as well, which is a great starting point. So you've captured imagery, you wait a few hours for the crosswalks to be generated, and then you can use that as a starting point to click on them, see which images they've been detected in, go to OpenStreetMap and add them in. And so here's an example of like the pixel segmentation
07:22
of the crosswalk itself. You can go in and add that where it didn't exist previously. Other really useful tools for mobile mapping include a Go Map. I really love this tool. I'm actually hoping to meet the developer if he's here at State of the Map US. Huge fanboy of his. This is Bing aerial imagery again, toggled.
07:42
This is the dog, I always struggle to say this, the bag that you get to pick up your dog feces. So you can add this when you're on the go with Go Map. Really useful for nodes, more so than Waze. Some people like to add Waze for pedestrian infrastructure when they're walking. I prefer to do that at home, but this is good for nodes,
08:01
things that might be related to pedestrian infrastructure. Vespucci, same sort of deal here, great for mobile mapping on Android. You also have StreetComplete. So once you've done all this, you've added like the base layer of coverage. You've maybe added some additional data with Mapillary or Go Map or Vespucci.
08:20
StreetComplete is great for quests, it's fun. You can do additional things like does it have tactile pavement? Is it lit at night? What's the surface type? All useful information. So StreetComplete is great and fun and apparently also really good for date night. This was a blog post that said
08:41
it was a really good night out with her partner. So keep that one up your sleeve. In terms of our display maps, particularly on Facebook, if you're thinking about a Facebook place, that's where you'll be able to see some of the pedestrian infrastructure you've added or that you've mapped out an open street map. Probably a bit hard to see here,
09:01
but that's a restaurant in Folsom near where I was living and you can kind of get a better idea of how to get there on foot. Mapbox Streets, again, kind of hard to see pedestrian infrastructure on their default layer, but if you zoom in, you can see it there. But really, what's the point of all this?
09:20
The point of all this, hopefully, is to actually get better pedestrian infrastructure. And the good thing about the city of Folsom is that they have a plan. Governments are good at that. The first one was done with a consultancy called Kimley Horn, and this consultancy put together the plan in 2014 that had a lot of goals around
09:40
people walking conveniently to their destination, feeling safe walking, facilities provided for people of all age groups, disabilities, so people can be more easily mobile, and then there's a huge commercial aspect to this. I think a lot of cities have proven that actually when you make a city easy to walk around,
10:01
people spend more money. They're more willing to just go from store to store rather than go back to their car and drive to the next place. And so it's really good to see that they have this plan. They actually just released the Active Transportation Plan, which is a plan for this spring. It's a PDF document, which you can go through and you can tell them exactly
10:21
where you want things to be improved, but it's not built around a map. And I think some of the previous talks showed how useful it is when you actually have a map as your central framework for discussion, particularly around emergency services. Same is true with government. They think around maps, and so if we can communicate to them using,
10:41
showing them what exists on the ground currently and what doesn't, we can help. So in terms of Folsom, it's about the size, it's slightly larger than Manhattan, obviously a lot less population, just to keep in mind in terms of density and moving from A to B. It is probably harder than Manhattan to get to the shops. Your local bodega, flat white,
11:01
is usually a drive away rather than a walk away. But downtown area is quite walkable, but there's a lot to be done just to make it easier to go to the grocery or to go get a bottle of wine or to go to the doctor. So all these things, unfortunately, are very drive centric. So here are a few quick maps before I wrap up.
11:21
This is the Folsom city government. What they've determined is important. So they've mapped out key pieces of infrastructure, mostly to get from A to B, signage that needs to be improved, and then some of the highest priority districts. There's still a lot that's missing here, and that's why I'm excited about OpenStreetMap as a tool to go out and map this
11:41
and to show them specifically where it needs improvement. So in terms of the steps, ID, Rapid, and JOSM for the base level of coverage, Mapillary street level imagery for the bits that are missing in between, GoMap for nodes, and Vespucci for nodes that might be related to pedestrian infrastructure,
12:02
and then lastly, StreetComplete for date night. So hopefully that's somewhat useful for you. I would love to chat more. I heard there's people who are really passionate about pedestrian infrastructure. I think there is like acknowledgement that this is an underfunded level in terms of the government's budget generally, and it's common around the world.
12:21
As I talked about last year, it's common that these things aren't funded as well as they should be. So thanks for listening. Thanks to the State of the Map organizers. If you want to connect with us, we have Map with AI Twitter handle. We have the Mapillary Twitter handle as well as a Facebook group, Mapillary Worldwide, which some of you are already in. You can connect with me personally here.
12:41
And looking forward to chatting offline about pedestrian infrastructure. Thank you.