How to use data in a better way and improve OSM


I really like OpenStreetMaps, and I have some ideas how to make it better. Ideas I cannot realize, 'cause I cannot code and I don’t know enough about these things. But probably somebody does like these ideas or parts of them and has the ability to realize them. That’s why I post them here.

The main idea is that OSM should gather data which is generated when people are using OSM. If somebody uses OSM for navigating (for example with Navit), there is a lot of data which doesn’t come back to OSM at the moment, but would be very helpful.


If Navit would have the option to automatically log your routes and submit them to OSM if online (again), OSM could be improved in different ways:

  1. Data could be used to realize which means of travel was used , for example bike, car, truck, etc. by the min/max/average speed. (See below to understand how this could be used)

  2. Automatically find new roads, when there are many reports about a way which isn’t already in the database.

  3. Automatically assume that roads are only one-way, if they are used only one-way. Here it could be helpful to distinguish between cars and bikes (see 1).

  4. Automatically calculate average speeds at different times over day / week. This data could be used to find the perfect route at any time. It would make sense to distinguish between different means of travel (see 1).

  5. If people use devices which have GSM and connect GSM-Information with GPS, OSM could be extended to build a map with the different cells & signal strengths on different positions. This way it would probably become possible to navigate without GPS.

  6. In a few years, when mobile internet becomes more widespread, this could be extended to not only fetch street-data from OSM in realtime, but also report and get traffic information in realtime. If, for example, a lot of cars (which use OSM) are stopping at one point they do not stop normally, we could guess that there is a traffic jam and reroute other users.

These are just some ideas, I suppose we could do a lot more things with this data.

Yes, there are privacy issues, I know. Probably this problem could be attentuated if the first and last $x minutes of routing are not reported. In every case there must be an option to turn logging off at all. But all in all I think the massive improvements which this data would make possible are more important than these issues.

What do you think?

Best regards,

No one of the devs is reading here, i suggest your create a ticket on

There are people who are trying to do this commercially since GPS navigation has become such a big thing last year. Haven’t heard any success stories yet…

Duhulefu, I understand that you are excited I an too, this would be one of the coolest projects that could be built on top of Openstreetmap… :slight_smile: Sadly it’s a very large project and the only thing in common with OSM is that you need a GPS for both things.

Yes, you’re right I am excited. :sunglasses:

But I disagree in one point - I think it has more in common with OSM than GPS! OSM is the perfect place for this; it i has an database already, it is already known and integrated to applications, users are more motivated to participate if you compare it to an commercial app and last but not least there are devs who knows a lot about this materia.

Yes I know this is a large project, but it can start with a simple step. Collecting data. Of course I’m not a developer, but I don’t think it is much work to build an API which allows apps to report GPS-traces automatically. Simply store them, you can do crazy things with this data later (and you need a little bit of data to start developing useful tools I think). This would be the first and simple step.

Second step is to implement the option of auto-tracing and reporting to the apps which uses OSM with GPS. Of course this isn’t something OSM can do (as far as I know OSM does not develop any apps) but OSM has to do the first step and write an API.

And I’m sure if this works there will be people whith bright ideas and the will to implement them. Think big, but cut it down to small and simple steps.

just thought I’d put my 2cent worth as I’ve seen a previous thread on the mailing list about automatically converting GPX tracks to “ways” , I think the short version is its not a simple change and its not without some risk too. as suggested I would post or search the “Talk” archives

I can’t answer it all at this moment. But just wanted to say; You can do a lot without developing things and that is come up with very simple tasks with small steps to realize your idea, you can of course not develop everything but you can make concept sketches of how things should work.

Perhaps buying ~100 GPS units and hand out to taxis in an uncharted area, have enough batteries for 3 days and then collect it all and have a map. I think this could be an interesting proof of concept for your idea.

What mtrax says is more or less what I think as well, it’s an wasy project when it’s done but until that moment it’s going to be hard. :slight_smile: We take Google for granted at this point, but before 1998 it didn’t exist.

On emj’s suggestion.

If I am not mistaken the London mappers were handed the gpx tracks of the pizza couriers. Fulfills some of the requirements: known vehicles, shortest routes, goes everywhere and alsways max speed (or higher).

The London mappers may want to share their experience.

You mean the old eCourier deal or something new?


correct, with this hint I found it back here: There it suggests that data have come in and can be used for tracking.

Brings me back to the start of this discussion: the map making is (as far as I know) still a human activity and the GPX is a very small part of the information. Think of access restrictions, bridges etc.

Well I still am a big proponent for a big GPX data dump, and access restriction and bridges are actually possible to discern with loads of GPXs from various sources. Though you might still need landsat satellite pictures.

So stop thinking of just one GPX trace, think millions… :slight_smile: