There is an interesting situation around pedestrian navigation and the open map community. OSM's pedestrian map data is already unique. Not a single supplier in the world is able to provide as many sidewalks, crossings, and paths as OpenStreetMap has. But at the same time, it is still very far from convenient industrial use of this data. As a result, OSM and its community will again have to pass the strength test.

Pedestrian Routing Issues

He is next to the car

The main problem with pedestrian routing is that it is not alone. OSM maps already have roads that are connected by a community to a good graph. But pedestrians cannot walk on most roads. Ideally, you need to close all roads for pedestrians and draw a separate road network with separate tags for pedestrians. But now this “ideal” is unattainable. Footpaths are indicated by disconnected islands in various parts of the map.

As a result, the community and manufacturers of navigation software have the following dilemma: in order to build walking routes in general, you have to use roads from the car graph. But it’s unambiguous to understand whether there is a sidewalk near a given road and whether it is marked on a map without special tags. Therefore, the algorithm can build a route along the roadway, next to the drawn sidewalk. To avoid this, manufacturers of navigation software have to come up with various heuristics that complicate the already complex and costly routing algorithms.

Novice cartographers, seeing a road on the map, do not think that it should be noted for pedestrian accessibility, separately draw sidewalks or connect the road with existing sidewalks. As a result, the pedestrian graph is developing much more slowly than the road graph. Unlike many other route data, pedestrian data is not allocated in a separate layer. They exist together with the master data of the map.

It requires cartographers more

Since we have already talked about cartographers, it can be noted that the graph requires more from the cartographers themselves. The usual intersection of two roads with sidewalks turns into a big roundabout.



Each underground passage turns into a set of at least two stairs, a tunnel and two connecting roads. A bunch of tags appear that aren’t displayed in editors or on standard OSM styles. Zebras, sidewalk, stairs. For example, the stairs have a direction (up - rise), but it is not drawn on the map, and many stairs are marked incorrectly. Agree, for a pedestrian there is a difference - to climb or descend. If you take into account the needs of "non-standard" pedestrians (people with scooters, prams and wheelchairs), then they need even more data, for example, about the presence of ramps and ramps. Although there are tags for all of this, few even know the experienced participants, and even fewer who place them.

Square

In general, pedestrians do not follow clearly defined lines. They have different areas where you can walk as you please. Although all the same, they go along the shortest routes, the area is drawn on the map separately. And the shortest hiking trails beneath it are often forgotten to indicate. Neither MAPS.ME nor any other application or engine can do routing by area. Moreover, the community is actively discussing the issue of the transfer of roads in the asphalt area with separately drawn markings. So the problem of areas may still come back to us in the road graph.

It’s worse on the pictures

To top it all off, pedestrian roads are almost impossible to draw from satellite imagery. All independent paths and roads that are not laid along highways are blocked by tree crowns and are not visible. Also, as some fences between the houses are not visible. Not to mention forest paths.

Destructiveness of minor edits

And mapping on a live graph is reminiscent of bomb disposal. Without special skills. All edits instantly go to a common database. No matter how we try to improve our routing algorithms, they are very susceptible to the quality of the graph by which the route is constructed. And I have already said that, along with sidewalks, each intersection is a difficult roundabout, and it is very easy to cut it with it. For example, forget to add common points to intersecting lines. So it turns out that during each data update you are waiting for where it broke this time.

Community Slow Change Management

Well, the overall slowness of a large cartographic community is all that completes. New tags come up hard. Proposals, the official proposals of the new notation systems, are not being completed. Initiatives rarely go beyond regional communities. To describe the scale of the problem for the uninitiated, I will give just one example: in OSM, there is still no mechanism for designating any areal objects. All areal on the map, including the continents themselves, are closed lines and conjectures of drawing programs.

Solution Options for OpenStreetMap

In general, if you look closely, you can understand that all of the above problems are not new separately. The community in one form or another has already met with them and somehow solved them. Therefore, I will describe some solutions that may possibly resolve the situation.

Run as is and wait for activists who will correct

The main OSM mechanism is time. Enthusiasts will use the first navigators, collect rakes and edit the map. The higher the data quality becomes, the more people will be willing to use the product. So, step by step, first the districts, then the big cities will reach the level of quality pedestrian counts. This week, MAPS.ME has officially started supporting hiking trails. You can appreciate all the benefits of pedestrian routing simply by downloading the application on your phone.

Facilitate the first entry for beginners. ID and Wizards

Mapbox does a tremendous job by developing tips during mapping. No matter how powerful JOSM is, but when a beginner is taken as a card, it is better to give him an easy way to find out what other tags you can fill out. And here all sorts of tips and wizards fell perfectly. Ideally, you should minimize the search for tag names in the OpenStreetMap wiki. As a bonus, this will ease the problem of typos in tags.

Mobile Mapping Applications

Continuing to simplify, mapping applications need to become more mobile. Let beginners be able to make easy edits directly from their mobile phones. We leave the process of collecting data to a bypass sheet and transferring them to the program for experienced community members. This is especially important for information that cannot be seen from the satellite. For example, information about stores.

Validators

The problem of data corruption is also not new for OpenStreetMap. The road graph is regularly broken, and validators help to cope with this. Take a look at projects like KeepRight and Peirce QA . So far, there are few such tools for the pedestrian graph, but the technology itself has proven itself well and can be used here. In addition, such sites are elements of gamification when making large-scale changes or maintaining the quality of maps. However, not all members of the open card community are aware of these sites.

conclusions

As you can see, none of these solutions can become a silver bullet for OSM. It takes a lot of strength to be able to take the best of every decision and move the map to a whole new level. However, the goal is grandiose. A pedestrian world graph is unprofitable for any mapping company. This is precisely the opportunity that can reinforce the triumph of open data over proprietary data.