Happy Birthday, OpenStreetMap(tomtom.com)
tomtom.com
Happy Birthday, OpenStreetMap
https://www.tomtom.com/blog/maps/tomtom-openstreetmaps-mapmetrics-map-editing/
42 comments
TomTom provides mapping data to Apple maps, Bing maps and they power many automotive in-dash navigation systems: https://www.tomtom.com/customers/
Very likely they use OpenStreetMaps to enrich their data and it's nice to see that they contribute back.
Very likely they use OpenStreetMaps to enrich their data and it's nice to see that they contribute back.
Worth noting that Steve Coast - the author of this post - was the founder of the OpenStreetMap project in 2004 (he joined TomTom in 2019).
Woo!
Happy to answer any questions...
Happy to answer any questions...
What does TT ultimately intend to do with OSM data? Do they intend to:
- provide OSM as a second option in their map products
- gradually switch over from proprietary data to OSM
- switch to OSM but only in select countries
- do something else entirely?
- provide OSM as a second option in their map products
- gradually switch over from proprietary data to OSM
- switch to OSM but only in select countries
- do something else entirely?
Where is this supported? Seems only a few selected areas?
The aim is globally, we're working on speeding it up and cleaning out trains and water... We'll probably finish the US within a few days and move on to the EU or other places.
Can imagine train traces particularly being tricky. Might have to build a separate routing graph of just train lines and filter out any traces that match that too strongly - though I don't think people mapping train lines in OSM work too hard to get train lines' routeability right...
Great minds think alike...
What is the business case for this? I don’t understand how TomTom has an advantage by improving openstreetmaps.
TomTom relies on maps for many of their products. OSM provides maps. Doesn't exactly sound like rocket science to me.
So far as I was aware, TomTom provides a commercial alternative to OSM data. Them helping OSM in that context takes some rocket calculus / is a valid question.
Their main products are centered on fast GPS tracking solutions, which works really well with maps and mapping out maps, but at the end of the day are "just" coordinates. Selling maps in and of themselves is not their core product.
Commoditize Your Complement
TomTom has GPS traces, but if those traces lead to OSM crowdsourced corrections without rolling a vehicle then they get better data faster. This may be particularly relevant for highway changes with no associated street addresses.
Many of the red areas I see on the map, particularly in France, are fairly obviously where someone's been running a TomTom while on a train or a boat. Of course that's not going match OSM roads. I'm a little surprised this is not mentioned in the blog post - it seems a bit obvious.
>"As you can see above it’s a work in progress: MapMetrics flags, in red, Colorado ski areas and the Denver Airport. Neither of these are bad per se, but ski run and taxiway routing isn’t a thing yet! Over time, MapMetrics will filter these out to focus on just the roads."
It's great to see corporations contributing to OSM. There's only a couple top contenders for mapping data, and the best way to compete with them is to collaborate on a shared resource.
It's a complicated relationship, since companies often struggle with respecting community norms and the overall goals of the project, but I tend to agree.
I find this a good way, where they expose their resources via tools that makes improving OpenStreetMap data with data they have access to easier. This isn't exposing their maps directly, so it does not come with their assumptions.
You can see why they might not want to share data they had to spent money to gather when their competitors would then get access for free.
In what way? I'm interested to know more about this.
It easily happens that employees do whatever it takes to get their internal tool to show the right thing, or use an automated tool, without ensuring they don't destroy or duplicate existing data. (made-up example: someone is refining the mapping of streets, but destroys information about adjacent/connected footpaths in the process because they don't matter for their delivery-vehicle use case. Or maps street types as they would in country A, but in country B a different norm has been established, so the edits don't fit the surrounding data)
Data quality can be a problem, where the company data source does not match reality (anymore). Or it's difficult for the community to verify your data and be convinced your quality is good, especially if there are other concerns about your edits.
OpenStreetMap operates on soft consensus and can have multiple acceptable ways of doing a thing. Mass-editing that converts one form to other is frowned upon unless wide agreement that one way is deprecated has been sought. Often related to the first point, when company tooling prefers one style or employees only have been taught about one method, but in an area the other mapping style is widely used.
Similarly, communication to resolve conflicts can be lacking: if another editor challenges what you are doing (because they think you made a mistake or are overstepping norms), you need to react and fairly consider their point. OSM has clear expectations that if you are doing coordinated edits you clearly identify and announce your work and respect feedback.
Data sources/licenses can be a concern: what you find acceptable to import into your company database might not be acceptable to include into OpenStreetMap, and how does the community tell that your data is good and won't get the project in trouble later?
Data quality can be a problem, where the company data source does not match reality (anymore). Or it's difficult for the community to verify your data and be convinced your quality is good, especially if there are other concerns about your edits.
OpenStreetMap operates on soft consensus and can have multiple acceptable ways of doing a thing. Mass-editing that converts one form to other is frowned upon unless wide agreement that one way is deprecated has been sought. Often related to the first point, when company tooling prefers one style or employees only have been taught about one method, but in an area the other mapping style is widely used.
Similarly, communication to resolve conflicts can be lacking: if another editor challenges what you are doing (because they think you made a mistake or are overstepping norms), you need to react and fairly consider their point. OSM has clear expectations that if you are doing coordinated edits you clearly identify and announce your work and respect feedback.
Data sources/licenses can be a concern: what you find acceptable to import into your company database might not be acceptable to include into OpenStreetMap, and how does the community tell that your data is good and won't get the project in trouble later?
E.g. by overwhelming a local community with edits from a swarm of outsourced contractors from a wholly different country. Good luck keeping up with verification of their work.
Or making mass edits / imports without discussing with local community beforehand.
Or as in the first point, but with "AI" as FB did.
Or making mass edits / imports without discussing with local community beforehand.
Or as in the first point, but with "AI" as FB did.
What did FB do on a OSS project with AI?
It seems that the MapMetrics data mentioned either isn't available everywhere or (less likely) their site is being crushed. The data only loads for California and Colorado.
Here too, it seems to be available for western Netherlands, parts of northwest Flanders, all of France, California, Colorado, Albania (fewer trace density it seems though). Maybe they're still importing/processing? Calculating routes for all over the world is a lot of data processing.
There is also a layer named OSM202106 (the default is OSM202103) with even less coverage (e.g. not France), but covering all of the Netherlands rather than just a part of it and covering the USA east coast.
There is also a layer named OSM202106 (the default is OSM202103) with even less coverage (e.g. not France), but covering all of the Netherlands rather than just a part of it and covering the USA east coast.
If I may be so bold, scrolling around the Netherlands and what is covered of Germany, it seems pretty useless.
- The resolution is so low, it's really hard to tell if there might be missing roads/routes or if this concerns water/rail.
- There are so many water/rail mistakes, there is red everywhere also in areas where I know every road in existence is on there. You have to evaluate each tile manually for all of the covered landmass.
- Just taking out air/rail/water might not solve the problem. Looking in residential areas with no water or rail nearby, there are still yellow tiles popping up at random with no indication as to why. Having the offending trace(s) would be really helpful here, but those are of course not shareable for privacy reasons. Perhaps short cutouts could be shared that are between 25 and 75% of the route (so not near the source or destination)? That would make it very clear if the person is on a bus (that obviously won't follow the shortest path in most cases) because the trace would go past the bus stops and often linger there, or if it was a cyclist for example.
As a frequent OSM contributor I love such initiatives and I'm very happy to see TomTom getting more involved. TomTom's map quality is so far behind OSM (globally), I was wondering who'd even still considering buying from them so it makes a lot of sense to combine forces instead. User data is the main advantage Google has over OSM, for both live traffic and purposes like these, so I'm very happy to see innovation here! I'll definitely be checking out updates to the site.
- The resolution is so low, it's really hard to tell if there might be missing roads/routes or if this concerns water/rail.
- There are so many water/rail mistakes, there is red everywhere also in areas where I know every road in existence is on there. You have to evaluate each tile manually for all of the covered landmass.
- Just taking out air/rail/water might not solve the problem. Looking in residential areas with no water or rail nearby, there are still yellow tiles popping up at random with no indication as to why. Having the offending trace(s) would be really helpful here, but those are of course not shareable for privacy reasons. Perhaps short cutouts could be shared that are between 25 and 75% of the route (so not near the source or destination)? That would make it very clear if the person is on a bus (that obviously won't follow the shortest path in most cases) because the trace would go past the bus stops and often linger there, or if it was a cyclist for example.
As a frequent OSM contributor I love such initiatives and I'm very happy to see TomTom getting more involved. TomTom's map quality is so far behind OSM (globally), I was wondering who'd even still considering buying from them so it makes a lot of sense to combine forces instead. User data is the main advantage Google has over OSM, for both live traffic and purposes like these, so I'm very happy to see innovation here! I'll definitely be checking out updates to the site.
Looking at the places I'm familiar with, there's also a pattern of places like high traffic intersections (probably lots of people avoiding those?) and places with lots of foot/bike traffic, where people don't travel along the shortest path but along the most beautiful or the one with the slowest elevation change.
I guess there's lots of useful information for improving routing algorithms in the raw data. I too am skeptical about its usefulness for mapping.
I guess there's lots of useful information for improving routing algorithms in the raw data. I too am skeptical about its usefulness for mapping.
This is cool data! ...however, I must ask --- where does the data come from?
HN certainly has a side that loves open-source data, but it also has a privacy-loving side. Focusing on the privacy aspects, has TomTom given any consideration to how it respects the privacy of its users/does it allow for opt-out of storing and analyzing its users' GPS traces?
HN certainly has a side that loves open-source data, but it also has a privacy-loving side. Focusing on the privacy aspects, has TomTom given any consideration to how it respects the privacy of its users/does it allow for opt-out of storing and analyzing its users' GPS traces?
Can anyone recommend good interface to OpenStreeMaps? I installed OsmAnd on my Android Radio (9" 720p) and find menu options/icons are too tiny (and too many) to operate while driving.
Surely warmly recommendable: OruxMaps.
I think you can find the apk at oruxmaps.com
> to OpenStreeMaps
Can I suggest you also try DerStefan's OpenTopoMap, at opentopomap.org .
Also, through the GitHub based instructions from DerStefan, you can build your own maps (tiles). It is an job requiring equipment and time, but it can lead to amazing results (well above the maps you can find around).
I think you can find the apk at oruxmaps.com
> to OpenStreeMaps
Can I suggest you also try DerStefan's OpenTopoMap, at opentopomap.org .
Also, through the GitHub based instructions from DerStefan, you can build your own maps (tiles). It is an job requiring equipment and time, but it can lead to amazing results (well above the maps you can find around).
How about Organic Maps? https://f-droid.org/packages/app.organicmaps
Thanks, I'll definitively give it a go.
Osmmaps + leaflet.js = luv
Some years ago you had to pay for updates and they were very slow to add new streets. Has this changed with OSM?
Depends on the package you buy. I have a "lifetime subscription", so theoretically I'll never need to pay for an update.
...but then newer devices only work with newer versions of the service, for which you haven't got a lifetime subscription. Eventually, your old TomTom stops working and your subscription is now worthless so you have to buy a new one.
...but then newer devices only work with newer versions of the service, for which you haven't got a lifetime subscription. Eventually, your old TomTom stops working and your subscription is now worthless so you have to buy a new one.
If you look at the data that the MyDrive smartphone app shows for each device, you'll notice that the "lifetime subscription" has a fixed expiry date. It appeared to be ~10 years for the devices I looked at. I could not see this information any other way except via the MyDrive app.
OpenStreetMap is a great product. It's such a shame that they've got licensing wrong. Now a lot of FAANG level companies can abuse it and sell free product with a bloat and trackers.
Congrats for your courage Tomtom! If I ever plan to buy a GPS or navigation device, I'll certainly consider your history.