Hacker News new | past | comments | ask | show | jobs | submit login
Quest-driven OpenStreetMap editing for Android (github.com/westnordost)
156 points by krzyk on May 10, 2017 | hide | past | favorite | 52 comments



I was running on some unmapped forest roads this weekend. When I used OsmAnd (which is my goto map app) I realized the roads where missing. I took a look at Google Maps via my mobile browser and realized the roads were there.

What is the easiest way to contribute and map out these roads on OpenStreetMap? Before I've edited OpenStreetMap by hand by using a satellite imagery overlay.

Could I use a mapper app like this to automate the process, or will I be manually editing GPX uploads in the end anyways?

EDIT: Had to try it now. Simple and polished. I don't see a way to map out new roads, which is understandable (I don't think it's in the scope of this super simple app). I had some trouble understanding why the inner city of my town had no "quests". It turns out you have to go to the application menu and manually "Scan for Quests here". Some places might seem to have few quests. However, zooming in a lot revealed plenty of quests for most any urban area. Kudos for making it available via F-droid!


For adding streets it would be best to use iD (web editor from osm.org) or JOSM desktop app.

If the satellite imagery is not available (iD one or in JOSM - those are approved satellite sources, that we can use - you can't use googles). In iD you just click on 3D hamburger on the right and select the source.

Other option is to make a gps trace you can use any application and upload GPX it later on on the osm.org webpage or you can use OSMTracker for Android and make a trace and upload it directly from the phone.


The issue with satellite imagery is that usually forest road/path are missing because they are covered with foliage. So I concur in this case, the easiest way is to create a gps trace you can upload and then edit with iD on openstreetmap.org


OSMTracker has handy points and text notes, which makes it my choice for detail mapping a blank area.


I'm using Locus Pro ( http://www.locusmap.eu/ ) for hiking, and Vespucci ( http://vespucci.io/ ) for map editing. Tried various Android editors; in the end, they're all clumsy compared to the powerful but computer-oriented mouse+keyboard interface of JOSM.


I second vespuci.

I usually use MapsWithMe a lot so I add POIs all the time.

For paths I just use Vespuci on mobile.


OsmAnd can record tracks, which you can upload to openstreetmap.org (in GPX format, which OsmAnd uses as well). Then using the iD editor (the built-in, online one) you can overlay that track.


I've found the editing interface on the website to be quite good, you can upload GPS traces if you don't want to map them out by hand.


> What is the easiest way to contribute and map out these roads on OpenStreetMap?

Dunno about easiest, but the most effective is probably to just buy the detailed satellite/aerial imagery, (http://microsites.digitalglobe.com/30cm/) run some machine learning algorithms to identify roads, and upload that. https://github.com/trailbehind/DeepOSM https://developmentseed.org/blog/2017/01/30/machine-learning...


Automated edits are discouraged in OSM, and if they are going to be done one has to send a notice to the appropriate mailing list and create the edits using additional username (so it will be easier to revert the changes).

See: http://wiki.openstreetmap.org/wiki/Automated_edits


I wouldn't say discouraged, just they have to be registered. See the list of all the imports, planned or finished: http://wiki.openstreetmap.org/wiki/Import/Catalogue. In a lot of places pretty much all of the data came from automated imports. People checking the map is good, but manual data entry is a pain and there aren't many doing that.

It looks though like Facebook beat me to the punch: http://wiki.openstreetmap.org/wiki/AI-Assisted_Road_Tracing


No to mention all the copyright potential problems.


That's why it's only possible to use whitelisted data sources: http://wiki.openstreetmap.org/wiki/Aerial_imagery . Given the scope of this task, "find a license-compatible imagery" is a non-issue.


run some machine learning algorithms to identify roads

If it was that easy we'd already be doing that. 30cm imagery is still very low resolution for tasks like spotting dirt tracks under foliage, plus even the current state of the art in image recognition gives far too many false positives and negatives to be really useful without having someone manually double checking everything.


FB Engineering is getting there, actually. https://code.facebook.com/posts/1676452492623525/connecting-... - of course, they're currently mapping buildings, which is an order of magnitude simpler task.

As far as "someone manually checking" - you'd need that in any case, but it is quicker than someone manually doing all the tracing work.


Facebook is working on roads too:

https://forum.openstreetmap.org/viewtopic.php?id=57942

They also had an earlier attempt that was pretty messy.

The identification of roads works pretty good. Separating roads out into main roads and secondary roads and access roads works less well.


You can't trace manually except from approved sources -- e.g. Google Maps is forbidden. I think this would fall in the same category.


Actually, as of yesterday, there is an approved source from DG. http://wiki.openstreetmap.org/wiki/DigitalGlobe

The actual issue for tracing paths is that they're hard to distinguish, even from hi-res aerial imagery; a bit of ground knowledge of the area helps a lot, so that you don't map nonexistent paths which are actually fences or such.


Great! Yahoo's maps are generally good for tracing, but the quality varies from place to place. Hopefully this improves the situation.


Have you already try OSM contributor? (https://github.com/jawg/osm-contributor)

I work for Jawg, the startup behind it and we are very close to OSM France.

Feel free to try it and ask questions about Jawg if you are interested in maps, tile-server or something else. (https://www.jawg.io/)


> However, zooming in a lot revealed plenty of quests for most any urban area.

Yeah, I think they could improve the UI by putting a single marker with a digit to indicate there's multiple quests to complete bundled together instead of displaying nothing at all.


Great to have you on board! Please do add surface tags when you're mapping forest roads - it makes all the difference for cyclists and other users.


Imo this app is a really good idea but the one of an automated difference check between google maps and osm road paths, at least for specific areas and using this app, would be really great and usefull to commit people mantaining osm even more!


No. Sorry, but it's not allowed to use Google's maps for anything OSM-related, manually or automatically. Google has its own ToS and it explicitly forbids reuse in other maps; it would be unwise to start pirating its map.


When the UK government released cut down versions of their OS mapping, then several diff tools that allowed you to see where roads differed between the two systems appeared shortly after. I'd guess it's only copyright issues stopping this happening with Google.

http://wiki.openstreetmap.orgwiki//OS_Locator#Comparison_too...


Here is the correct link: http://wiki.openstreetmap.org/wiki/OS_Locator#Comparison_too...

Didn't know it wasn't allowed to retrieve information from google maps and compare it to other maps


You could, I guess, compare maps for your own use (that's their purpose, after all); but what you must not do is transfer knowledge from Google's map into OSM (or vice versa).

Would "I looked at GMap/OSM comparison and then reused what I saw in OSM" be wrong in the context of law? Would that be considered copying from Google? IANAL, but it seems to me that Google's position would be stronger than yours - they do have a "no derivatives without written permission" clause in the ToS, and an automated check looks as a completely different intent than normal map use.


I'm not clear what you mean by "reused what I saw in OSM". You probably can get by with comparing google maps to OSM, but if there is any difference you need to physically go to the error location yourself and see what is really there.

It is common for map makers to intentionally put a few errors on their maps, if those errors appear in someone else's map that is proof of a copyright violation. The location and names of roads is a fact that cannot by copyrighted, but errors are a creative work that is subject to copyright. I first heard about this in the 1980's when map meant paper, I suspect google is doing the same thing in some form.


Okay, let me be completely clear: I think that even a comparison tool between license-incompatible maps is a bad idea. Better?

(As for "why it's a bad idea" - it could tempt people to copy data from one to the other, which is definitely a Really Bad Idea)


When contributing to OSM, it's best to not use any other maps while you work. Consider it a form of clean-room mapping, if you will. Editors like iD and JOSM already provide powerful access to approved imagery sources and GPS traces. There's no need to 'steal' information from commercial maps like Google.


That's what I'm saying. You are probably trying to reply to parent.


I wasn't suggesting to automatically copy infos from gmap to osm but to provide something like a notification that the two maps differ in a place/road.

So, i understand that i can't create an automated list that lists what are the differences between gmaps and osm.


I would advise against comparing, too. A good lawyer could prove that diffs between the maps are made from Google's content. In short: unless it's explicitly whitelisted (e.g. Bing Aerial is on the whitelist, Bing Maps is not), avoid.


Nice :) I used osm a lot in Germany, far better data than Google. But currently I'm in South Africa and the data for osm is usually outdated or incorrect. This might help me help osm :)


The inconsistency of OSM coverage from one place to another can be a bit baffling. While cycling around South Africa four years ago, I added quite a lot of detail to areas on the Garden Route. It’s a bit disappointing to look at the map and see that almost nothing has been done with these particular locations since then, even though it’s a heavily touristed area and a developed country where you’d think at least some locals would be interested in mapping. And yet, at the same time there are some rural areas around the world that seem out of the way, but they show continual improvements by successive visitors.


And yet, at the same time there are some rural areas around the world that seem out of the way, but they show continual improvements by successive visitors.

Those rural areas probably don't have any other source of good mapping data so there is more incentive of people to map them. Popular areas with lots of tourists probably already have good maps from other sources so people don't feel the same need to create new maps over them. At least that seems a plausible theory.

I know that several NGOs and humanitarian aid groups use and maintain OSM as their primary source of maps for some of the remote areas they operate in.


I'm living in the suburbs of Pretoria, OSM showed a Woolworth store that according to the security guy hasn't been there in 3+ years :D


Pretoria was mostly mapped by a single mapper very early in the project. There wasn't high quality aerial imagery available when he literally cycled every street in Pretoria.

Because the map "looks complete" Pretoria hasn't really attracted many mappers since his departure.

Cape Town has a much more vibrant community of mappers. I've also been able to get crazy high resolution aerial imagery for CT from the municipality which helps: http://coct.aerial.openstreetmap.org.za/


It's a wiki. You are invited to edit it! :)


I’m reluctant to encourage people to edit OSM just like that. OSM isn’t any more a wiki than Wikipedia is nowadays; that is, while it can be edited by the community, there are an enormous amount of rules and caveats that contributors should be aware of before they start editing, and it may well be that their edits will be reverted if they aren’t clued up to how things are done.

In my own neck of the woods, there’s an OSM group with frequent meetups to guide newbies through editing. I think the best way for interested people to get involved is by learning in the company of experienced mappers, not just all by themselves at home. If such a stringent perspective scares some people away, at least the contributions of those who do stick around will be much higher-quality.


> there are an enormous amount of rules and caveats

There really aren't. OSM is very freeform, almost anarchistic, compared to Wikipedia; very few edits are reverted and very few editors are banned.

I'd say there are probably five rules:

1. Don't copy from other maps

2. Edits should be 'verifiable' and 'on the ground' - i.e. no subjective information

3. Be liberal in what you add, conservative in what you delete - someone else is interested in it even if you're not

4. There's a special process for imports and automated edits

5. Be prepared to engage with other contributors


If people are reverting reasonably accurate contributions, that's the problem to fix.

There are well mapped areas where just barging in and changing things will cause some issues, but for most of the planet the bar for improving things is quite low.


I strongly disagree that the bar is low. It’s not at all unusual for people who quickly sign up for OSM to remove a defunct POI, to accidentally remove other nodes along with it because they don’t really understand yet how editing is done. It’s also common for newbies to trace from wildly unaligned aerial imagery, or to use the wrong tags because they don’t know to consult the wiki any time they use a new tag. OSM is years past the point where any additions are welcome, and like with Wikipedia, no edits at all are often preferable to low-quality ones.


The response should be to encourage them to slow down a little bit not to chase them away entirely.


It's why I welcomed the app, it lowers the barrier of entry :)


Maybe cluster objects at lower zoom levels? As it is now it's difficult to get an overview because of the quests only showing up when you're zoomed all the way in.


I think the gamification of contributions is extremely interesting and this is a great step in the right direction.

However - there needs to be more "game". I've personally witnessed the obsessive nature of GPS-based mobile games (Ingress mainly). If we could harness some of that towards OSM it could be amazing.

Considering the amount of energy gamers put into customizing and modding - if the right game harnessed OSM data then you'd have a willing workforce on tap.


That looks impressively more modern than my preferred OSMtracker, which is really just a GPS trace with notes. I'll have to give it a try.


I've done some quests this morning - and took a look at the map history around me to see many other people also doing edits from this app. This is great! The level of participation this is enabling is really impressive!


Does anyone know of a similar app for iOS? I'm 'addicted' to OSM and such small quests would be awesome to kill some time at bus stops or during walks.


Well, there's Maps.Me, which has some editing capabilities - I hear it's useful for entering e.g. opening hours or contact details for a POI (phone, website, etc.). I don't think it supports editing everything that's in OSM (only tried it briefly), but it could fit your use case; give it a try.

Edit: Apparently other editors exist for iOS: http://wiki.openstreetmap.org/wiki/Apple_iOS#Editing_the_map


I know and use Maps.me a lot, but it lacks the survey character which I specifically find intriguing in this app. Also, there's http://www.pushpinosm.org/, but it will probably cease to function in iOS 11...




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: