Open US GPS Data?

tobiasly writes "I read an article today about a map error on the popular Garmin GPS devices which often leads to truckers in a particular town becoming trapped. From my own experience, every electronic map I've ever seen (Google, Mapquest, my Mio GPS) has the layout of my neighborhood completely and frustratingly wrong. A quick search turned up only one open-source mapping project, but it's for New Zealand only. Why are there no comparable projects in the U.S. or elsewhere? Obviously such a project would need a good peer-review/moderation/trust system but I'd gladly put in the time necessary to drive around town with my GPS in "tracking" mode, then upload, tag, and verify my local data. Has anyone with more technical knowledge in maps and auto-routing looked more into this? Are there technical limitations to such a project? Should the government subsidize a project to create open, free, up-to-date electronic maps? Surely there is a public benefit available from such a project."

TomTom MapShare

[autophile]

Try TomTom MapShare [clubtomtom.com].

open street map?

[Anonymous Coward]

http://www.openstreetmap.org/ [openstreetmap.org]

for Argentina...

[hjf]

For Argentina, there is www.proyectomapear.com.ar

Re:open street map?

[morgan_greywolf]

I've seen this and it lacks a WHOLE lot of data. It will take an army of volunteers dwarfing the number working on even high profile projects like the Linux kernel to ever get this thing off the ground. Can it be done? Only time will tell, I suppose, but this project is lllloooooonnnnggg ways off from being useful everywhere.

Re:Didn't look very hard

[juanfe]

Open Street Map [openstreetmap.org] has a good and growing base of data for the US. Plus they link in to open source or freeware applications that you can load on PDAs, GPS-enabled cell phones, laptops, etc to begin creating traces that can then be turned into map data.

Combine that with Open Source GIS software [opensourcegis.org] to query the data source and you're in business.

For this to work, you have to have a huge pool of people willing to drive a lot. Even the big map players (NAVTEQ and TeleAtlas) have problems keeping data up to speed, and they have an army of people driving around double-checking existing street grids.

Re:TomTom MapShare

[Laughing Pigeon]

Try TomTom MapShare.

Unfortunately, this has nothing to do with an "open source project". It is more like:

1. Make something that is so-so.

2. Profit!

3. Let the people who pay a lot of money for this so-so product do work for You without paying them for it. These users will take Your product from the so-so stadium and turn it into a good product.

4. Even more Profit! without any costs.

Reminds me a bit of cddb... What the OP wants is something like Freedb.

slow to update too

[Anonymous Coward]

The current solutions are slow to update as well. I the road I live on now has been here for 7 years, yet it's a field with no streets according to google, map quest, yahoo, etc. People can't search for my address on those sites as it doesn't exist on there.

Tracks4Australia

[shogun]

In Australia there is the Tracks4Australia [gpsoz.com.au] project which uses user contributed GPS track logs to generate rural and remote area trail and road maps, mostly useful for 4WDers etc. They are working on a commercial product now but the basic mapset appears to be staying free.

Maemo Mapper!

[Alizarin Erythrosin]

If you use one of the Nokia internet tablets, try Maemo Mapper [maemo.org].

Re:Frustrating

[Kagura]

Unfortunately the same bad data keeps getting recycled everywhere, because companies are too lazy to verify things.

I think you are underestimating just how many roads there are in the US.

Source: National Highway System (United States) [wikipedia.org]
The National Highway System (NHS) of the United States comprises approximately 160,000 miles (256,000 kilometers) of roadway, including the Interstate Highway System as well as other roads, which are important to the nation's economy, defense, and mobility.

Further down in the same article:
The 160,000 miles of NHS include only 4% of the nation's roads, but they carry more than 40% of all highway traffic, 75% of heavy truck traffic, and 90% of tourist traffic.

That's a lot of roads. Stupid lazy companies... :)

Re:Government Maps - of course

[Azure Khan]

*cough*google maps street view*cough**cough*

Mapshare?

[Zerbey]

My TomTom device has mapshare built in, I'd be astonished if Garmin did not. I've made dozens of map corrections (mostly silly stuff like incorrect street names) and they seem to update the maps often. My neighbourhood has been around for a while so no problems with the street layout here. I believe TomTom use Teradata maps whereas most other GPS systems use a different company.

I would love to see an open mapping project though.

Re:That's a pretty big job

[fireboy1919]

I work at a GIS company.
Keep in mind that there's USGS [usgs.gov], and that's not the only source of public maps (though that particular source isn't really focused on making navigation easier).

Most states are now working on providing a unified system for people to put their map info into (currently the best source of maps is counties and property appraisers - both of which can easily be mandated to upload their data if it doesn't cost them much).

So give it time. In the US this will become something provided as a government service, and the only people selling things will be the ones writing software that analyzes the data.

Re:Government Maps - of course

[Trailer Trash]

Yes, and TIGER is put together by the USGS, and it already *is* the "open source" data that the geniuses here are talking about. If you find an error, alert the USGS. I've done it myself - call their number and ask.

Now, as for the fantasy of people driving around with a gps attached to their car (ha ha, isn't that stupid!), um, oh:

http://www.navteq.com/about/whatis_difference.html [navteq.com]

"NAVTEQ digital map data is built on the roads of the world. Over seven hundred NAVTEQ field researchers from approximately 168 offices drive millions of kilometers of the road network each year. To provide uniformity and maximize precision each team works to a single global specification. And each team has state-of-the-art equipment, including our proprietary GPS-based collection technology and GWS software.

These field teams are constantly verifying and updating the database, not only in terms of road geometry, but also in details. Each team finds and records up to 260 attributes--everything from addresses and road signs to turn restrictions--for each segment of road. The result is the NAVTEQ difference: digital map data that is precise, robust and multifaceted."

There's no pretense; Navteq has people driving around, with gps's, verifying speed limits, road conditions, etc. That's why companies like Google and Yahoo buy their data. Before you act like an ass, you might want to do some rudimentary fact-checking...

Check your local streets dept.

[kabocox]

You can blame the government mainly your local streets dept for this. I've noticed state and federal highways being much more accurate than local or rural streets. May your deity help you if you live in a town that likes to rename side streets every few months.

Sure, it would be nice if there was some federal D.O.T. streets db for the entire country that your local streets department could upload all their changes into and all the GPS map folks would just that. I doubt it'll ever be that clean cut or that your local street department will want to even give any other city much less state or federal government department access to updated street info. This is just my personal experience working in a city police department and occasionally trying to get this information from the city entities that physically make and should be tracking these things.

The more that I see that its difficult or impossible for intercity departments to communicate I tend to think that the only real solution is for Pizza companies or UPS/FedEx to partner with Google streets to actually physically map out where their fleets move through.

If your city has a GIS department, then that should be keeping track of this information.... You could always do a FOIA request for any arcview street centerline data.
The problem is that most of us have problems getting that "updated" arcview street centerline data into our lowest price GPS device.

Open Source UK GPS Data

[killthebunny]

We have been collecting GPS positions at 10 second intervals since we began operations in London in 2004 (we're a courier company with a technology twist). We have collected 173 million positions on a 24/7 basis (growing by about 1 million per day) across our bicycle, motorbike, and van fleet. We have been donating to OpenStreetMap for years and have released our data for noncommercial use via a public API http://api.ecourier.co.uk/ [ecourier.co.uk] under a CC license. Have fun!

Re:open street map?

[TheRaven64]

Being able to distinguish a bike path from a highway is significant

Are you looking at the same OpenStreetMap as me? I just looked up the area around my house on OSM and Google Maps. OSM has more accurate mapping of the extents of the park (Google Maps is just plain wrong here). It also shows footpaths through the park (as dotted lines - Google doesn't show them at all) and indicates the different road types correctly (Google uses nonstandard colouring for roads) and shows roads inside the university campus, where Google just shows a grey blob. OSM also shows the hospitals and carparks correctly (sadly not the pubs). When it comes to road names, both have some that the other lacks (neither has complete coverage, but both have all of the major ones).

Re:open street map?

[Anonymous Coward]

Point is, in OpenStreetMap you *can* fix errors.
If you are referring to USA data it's a direct import from the TIGER data set, which does need cleanup. Europe is much better (where there is coverage, but that's just a matter of time, really.)

Fixing errors on mapping sites

[halfabee]

Please forgive the slightly off-topic post...

Two of the biggest map data providers are Navteq [navteq.com] and TeleAtlas [teleatlas.com]. Each company has a section on their website where you can report errors in their maps.

Since they will need to review your submission and mapping sites like Google Maps and Mapquest only update their map data a couple times a year, it will be a while before your correction goes public (if ever).

No? There are commercial applications...

[Lord Satri]

I disagree. OSM is very useful in many areas, including where it is hard to find maps (try Baghdad [openstreetmap.org] for example). With the recent addition of TIGER data [slashgeo.org] for the whole U.S., OSM became useful even in the U.S.

this project is lllloooooonnnnggg ways off from being useful everywhere

This is obviously not true when considering there have been commercial applications of OSM for a long time [slashgeo.org] (Isle of Wight - October 2006). See also this related wrap-up entry [slashgeo.org].

I am amongst the ones who believe we're only seeing the beginning of OSM everywhere. Contrary to your comment, I believe it is happening and will not take that long to reach some level of overall maturity. As to why is doesn't need an army of volunteers? Because, as done with the TIGER dataset, datasets are directly piped into OSM, as done in the Netherlands last year [slashgeo.org].

Re:open street map?

[Anonymous Coward]

Have you actually looked into OSM? It's data attribution scheme is significantly more flexible than 'regular' GIS. It is not only able to distinguish between a bike path and a highway but also able to specify that the bike path is private, goes uphill, that horses are allowed and that the pub halfway is closed on sunday.

Prior art

[goodmanj]

Should the government subsidize a project to create open, free, up-to-date electronic maps? Surely there is a public benefit available from such a project."

This is a great idea. We could have some federal government institution which deals with lots of maps anyway take the initiative and create digitized map data for the whole country, using information from USGS quads. For "fact checking", they could mail out the map data to every municipality in the country, who would make corrections which would be incorporated into the system. The data would be publicly available from the government for free, to be used by open-source or commercial makers of maps and map tools.

Congrats! You've just re-invented TIGER, run by the U.S. Census Bureau. If you use map software, it probably uses TIGER data. If the data in your town is inaccurate, it's because your local government sucks.

Surveying (CAD and GIS)

[esocid]

I worked as a surveyor for a private engineering firm a few years back and it isn't a simple task just to collect data and upload it. This applies for GPS data as well that you have to upload into GIS, or the like, software and manipulate it with any data-correction and overlays to aerial or satellite photography. Trust me, I spent hours cleaning up collection points and trying to get it to match up with the overlays with GPS data for invasive species management plans for a national park I worked at using ArcGIS (which is absolutely terrible to work with in comparison to ArcView). The surveying part usually requires some sort of CAD to properly map out what information you have collected during surveying and in-the-field math to figure out what goes where. It's not as simple as you think it might be.

Re:Frustrating

[rdawson]

I had cartography friend tell me that often map errors were introduced intentionally as a form of copyright. A mapmaker inserts a bogus item, street, landmark etc. into the map as a watermark to detect copies of his work.

TigerData et al

[chelanfarsight]

1. I use TigerData as a GIS professional and frankly its often crap. It was a good start for a rushed product in order to launch a project, but I would not now nor would I ever rely on its accuracy without checking it. The TigerData for my area regularly has roads going off the sides of mountains, roads where there have never been roads, etc. Also, the TigerData for my area has not been updated since it was released almost 8 years ago.
2. As for "driving around" it would depend upon how accurate the device is. The local utility company I work closely with spent 5,000$ just on the handheld to receive subcentimeter readings and about 20,000$ on the base station to accompany it. Your typical yellow DeLorme unit is great for driving around but it is not a data collection unit I would use when building maps. Depending upon satellite coverage for your area (weather, tree cover, geography, the placement of the 3 satellites needed to position accurately) your store bought unit could be as much as 100ft or more off your actual location and rarely closer than 5ft. Again depending upon coverage and the device. Then add the need for regular updates and mapping changes.
3. An open source mapping project would be great, but it is currently rather expensive to actually collect and process the data needed to build accurate maps. A terrific source of addressing and centerline information is your local E911 Board. At least in my part of the world they do much of the fire district, centerline, and, of course, addressing for mapping.

Re:That's a pretty big job

[NoPhD]

Check out this website.

http://freegeographytools.com/2007/converting-us-census-tiger-data-into-shapefiles-for-free [freegeographytools.com]

Re:open street map?

[ageforce_]

I could help that project by uploading my route tracks but what if I use mapsource (garmin software) to look up the road name am I infringing on something?

Yes. Unfortunately you are not allowed to do that. Map-vendors are protecting themselves against copying by deliberately introducing errors. See for example http://wiki.openstreetmap.org/index.php/Copyright_Easter_Eggs [openstreetmap.org]

Re:open street map?

[budgenator]

Street names in the US are assigned by the government and the government can't own a copyright in the US. All most all of those maps have errors and they have a lot of the same errors because they are based on the TIGER, Topologically Integrated Geographic Encoding and Referencing system, database. The TIGER database [census.gov] is maintained by the U.S. Census and while they are huge, you can have a lot of fun with them especially when you mix in the Perl module Geo::Coder::US [cpan.org] and GMT, Generic Mapping Tool [hawaii.edu]. The TIGER is a database of any known and and a huge number of interpolated data points, for example my house is listed as a known point with it's "official" latitude and longitude, two blocks down is another known point and every house in between is estimated. One thing you quickly notice when playing with the database is that roads often have multiple "official" name, Roads may "officialy" exist but not physically exists and roads may physically exist with out "officially" existing. Roads can even meander and move, especially dirt fire-roads and trails in the woods.

Depends on the GPS you buy

[EmagGeek]

Map data will make or break a GPS product. To me, you are committing monetary suicide if you buy a GPS where you cannot regularly update the maps, and/or the manufacturer does not provide such updates.

Most consumer-level GPSes do not have updateable maps, and those that do just update you with a year-old map they got for cheap.

I have a Garmin GPSMAP-496 and I *love* it.

If you want a good GPS with an accurate map, you have to pay for it. The $100 Mio piece-o-shit GPS is going to have, at best, a 3-5 year old map on it that they picked up a license for on the cheap. I tried a Mio, and it didn't even have a new map for the intersection of North Wales and Morris roads in 19446, which had been redesigned a decade or more ago. The result: "Turn right down this road that doesn't exist anymore."

There were also many cases where it would tell me to drive a mile or more out of my way, only to turn around and go back. It also sent me down dead-end streets SEVERAL times because it thought they still went through. Again, these changes around town were made a decade or more ago, but the Mio had no idea because the manufacturer used really old map data.

Of course the map is wrong...

[Frugatti]

Of course your map is wrong; you don't live on an Interstate. A few points to make on the digital mapping companies responsible for all the maps, update cycles and TIGER. And some bias, I was a former employee of one of the digital mapping companies. In the US (And globally) There are two companies that are responsible for all the digital maps Tele Atlas (Owned by TomTom) and Navteq (Owned by Nokia). If you look at any GPS device or online map site you will see a copyright from one or both companies. Their business is driven by getting people from point A to Point B the fastest i.e. by routing you to nearest highway and having you drive on it for the longest amount of time possible. The main focus is the US highway system: Interstates, US Highways, State Highways, Regional, County and Municipal Routes and the major metropolitan areas. If you don't live in the metro areas there is very little business need for correcting the errors in your locality. Why fix the streets in Stowe, VT (pop 6,000) when many more people will be served if the streets and addresses were updated in Cary, NC (pop 130,000 metro area of Raleigh, NC)? Each company works on a quarterly update cycle where a new version of their mapping database is available for purchase every 90 days or so. Some customers get the quarterly updates some get annual updates. The GPS units and online mapping sites are only as good as the currency of their maps. Make sure you update you maps every time an update comes out. There is always construction and changes in the road system and old maps will not reflect the newer changes. Just because Google Maps says the copyright is 2008 doesn't mean the map has been updated recently. I know when I was working for the mapping company we were working on a huge project for a car company that would be taking the mapping database produced in Fall 06 and using it for the navigation systems in their 2008 cars. If you do have a specific problem go right to the source to get it fixed Navteq map reporter or Tele Atlas Map Insight. The US Gov't does have a free nationwide map you can use TIGER (Topologically Integrated Geographic Encoding and Referencing) Produced by the Census Bureau. When was the last Census done? 8 years ago? Yea that gives you an idea of how accurate the TIGER map is. TIGER was made for government applications showing very accurate municipal boundaries and topology of streets. To get an open source Map project going you will need a good sized server, Volunteers in just about every municipality, good database software that can hold every thing you ever wanted to know about streets (name, address, one way, truck and vehicle restrictions, routing info, Points of Interest, zip code, real time traffic data, gated communities, municipal boundaries, state locality, ect...) a great set of possionally accurate aerially imagery (preferable 10m acct or less) for alignment of streets, and did I mention a large army of helpers in every municipality. Just make sure you get the newest maps updates for your navigation device and go directly to the source for map fixes: Navteq map reporter or Tele Atlas Map Insight.

Re:Frustrating

[curmudgeous]

...Getting the facts wrong just to protect your copyright?...

This used to be common practice among reference work publishers (i.e. encyclopedias). They would insert the occasional useless, bogus information that the normal person would never have reason to read as a guard against plagiarism. If it showed up in a competitor's work then it was a clear sign that it had been copied, not researched. Map publishers have most likely followed the same practice assuming that the average driver would be smart enough to recognize a road problem before getting into too much trouble.

Silly publishers.

Re:That's a pretty big job

[Wannabe Code Monkey]

The only two suppliers of nav map data in North America are Navteq and TeleAtlas...

It looks like Massachusetts gives this data away for free [state.ma.us]. I found that page as a reference from a Wikipedia article about some state route in Massachusetts. The data looks to be very detailed, the dataset is around 100MB. Heck, just read the Road Inventory Data Dictionary [state.ma.us] to get an idea of what they record. And yes, I know it's in an Access database, but it wouldn't be that hard to translate into whatever format one would need.

They're called tags

[TobascoKid]

OSM already has those features - they're called tags (not layers)

http://wiki.openstreetmap.org/index.php/Map_Features [openstreetmap.org]

Re:open street map?

[Andrew Allan]

Bike paths? We go one better - we can make dedicated cycle maps based on OpenStreetMap data since we can put anything we like in the database, and render our own tiles using any cartography we can think of. See for example http://www.gravitystorm.co.uk/osm/ [gravitystorm.co.uk] - zoom in on London to have a look at cycle networks, bike shops, contours and all manner of customisations for cyclists.

There's so much more potential to OpenStreetMap than just what's on the front page of the website.

There's plenty of data out there

[BlindJesse]

I made GIS maps for years, always using public data. And there is a lot out there, and it is government subsidized, and it can be of marginal quality. The only nationwide project I am aware of is the TIGER project [census.gov], which is supposed to release a new, provisional data set every year. When you can get it to work, its pretty good. Federal agencies also often release their own datasets, and we would often have forest service, national park service, and blm data on the same maps, sometimes in overlapping areas. Then there are the county datasets. And the city where I live put out their own dataset a few years back. So there's plenty of data, and it is almost all free. Companies that charge for it often have done post-processing or packaging which I believe they have government contracts to do and are allowed to recoup their investment.

Where trouble often comes in is in projections, spheroids, datums and the like. GIS data on different scale will use a different model of the globe to pinpoint places, will use different coordinate conventions, often related to the agency that produced the set (eg the city always used something called state plane, the fs always uses the nad27 datum, well, mostly). Two datasets that have location information for the same road can be meters off simple because they are not represented in the same projection correctly either by the software or the person doing the projection. And these are just location issues. Tabular data is a whole other thing.

The poster sounds a little uninformed about GIS in general.

Re:Sometimes they're easy to spot...

[TooMuchToDo]

Link for the lazy:

http://maps.google.com/maps?f=q&hl=en&geocode=&q=Dummy1456&sll=37.0625,-95.677068&sspn=32.80241,75.146484&ie=UTF8&ll=52.784492,-1.615934&spn=0.012225,0.036693&z=15 [google.com]

Re:open street map?

[Richard Fairhurst]

Sure, you can import traces directly to the map.

Upload your trace, click "edit" next to it, and it'll open in Potlatch (the Flash-based editor - disclaimer, I wrote it :) ). Wait for it to appear then click the "Track" button to convert it to vectors that you can tag up, split and otherwise edit. It even runs a simplification algorithm (Douglas-Peucker) over the track so that you don't upload too many intermediate points.

Re:open street map?

[Matthew Bafford]

Have you actually looked into OSM? It's data attribution scheme is significantly more flexible than 'regular' GIS. It is not only able to distinguish between a bike path and a highway but also able to specify that the bike path is private, goes uphill, that horses are allowed and that the pub halfway is closed on sunday.

That doesn't even make sense. GIS [wikipedia.org] doesn't define any data structures (excepting in the general sense that spatial data is involved somewhere). There are some standard and common structures for storing transportation data, sure, but those aren't "GIS" any more than saying any given contact data structure is "database". The big vendors (Tele Atlas and NAVTEQ) actually include a good bit of useful information if you pay for it and more stuff is being collected all the time. They might not have the esoteric information about which pubs serve horses on Tuesdays before 2pm, though, but there's nothing about "GIS" that is stopping them from supplying that if they wanted to.

I've not seen the path data that either of the vendors supply (I work with road data more), but it'd surprise me if they didn't have classes of trails as well as use limitations. Elevation and grade is likely stored as well (and can be calculated or approximated from the elevation data if present).

As for hours of operation, the POI information I have from NAVTEQ doesn't have that information, but we didn't pay for detailed info. There's definitely nothing stopping them from providing it if they have it.

Re:TomTom MapShare

[Lumpy]

Free Map data source? easy. grab the Us census bureau tiger line Census data. That's what most of them out there use to begin with anyways. Last time I checked it was a free download. i was working on a linux GPS navigation app with it for a car pc project. Then I found how to run a windows app for it under wine and my need was filled.

The data is actually easy to parse. far easier than navteq data.

Re:open street map?

[Lumpy]

Tiger is still around....

http://www.census.gov/geo/www/tiger/index.html [census.gov]

2006 is last year though of the old data format. 2008 release will be a completely new format with better polygon definition rules.

The manufacturers don't make the maps

[Colin Smith]

They buy them.

You can buy them too. Very expensive, and most of the US maps are completely shite anyway. If you were to create a good database of routes, streets etc, the PND and phone manufacturers would love you.

The only really updatable maps I've come across are Google maps, and of course Nokia have Ovi on the way, where the whole point is to be able to sync routes/locations with your friends.

 

Re:Of course the map is wrong...

[theskipper]

The US Gov't does have a free nationwide map you can use TIGER (Topologically Integrated Geographic Encoding and Referencing) Produced by the Census Bureau. When was the last Census done? 8 years ago? Yea that gives you an idea of how accurate the TIGER map is.

In 2002 the Census Bureau contracted Harris to update the centerlines and attributes nationwide. Approximately 1200 of 3200 counties in the US have been completed with another 300 or so due in March. Details on the "MAF/TIGER Accuracy Improvement Project" are here: http://www.census.gov/geo/www/tiger/index.html [census.gov]

Re:The dancing dog observation

[babbling]

Hi ..bruce..,

Do you consider the software or the database to be unfeasible? You talk about scaling things up, but there's very few reasons why software that works in one neighbourhood would fail in another except for deficiencies in the data.

The idea behind projects such as OpenStreetMap [openstreetmap.org] is to build the data, using contributors who are local to the area that they are mapping. I think OpenStreetMap is only beginning to pick up pace, and it is already getting quite good considering that it has been quite a low-profile project until recently.

The software side of this (as far as I know) doesn't exist yet, but when it does get started, you might claim to be correct if it happens to be crude at first. Free Software projects often are because they tend to release earlier than commercially-driven projects would. One strength of Free Software is that it can never go bankrupt. We can refine our poor software until it is great without having to worry about a project running out of money. If the Free Software for GPS navigation is crude at first, it will only ever improve. Eventually, if people keep working on it (and they will, because someone in the community will be unsatisfied until we have such software), it will be good.