Creating our own GPS maps

Shotty · 07-11-2005, 03:30 AM

Maybe it is just that i've been reading too much Richard Stallman articles, but the fact that we have to pay (significant) amounts of money for maps of our publically created road system got me thinking.

Granted, the large companies have gone through the effort of creating their database. But what if we could create our own!

Imagine this.. a small app, and all it does it take in GPS co-ordinates (and perhaps other details like time and that) from your in-car GPS unit. When you are connected to the net, it uploads your travels. A (admittedly complex) server based system would analyse your traffic, comparing it with other travellers who also travel the same locations.

For example, it would realise that you've stayed in the same spot on a road for.. say... 1 minutes. It might hipothesize that this is a red light. This could be backed up but another user having the exact same phenomenon at the same location. Enough information thrown at it, and with some volunteers putting in street names, etc... we could have a pretty good shot at it!

The result would be a global map system in a standard format for use in an number of home-brew GPS navigation programs, etc...

Anywho, i've had this idea running around in my head for a few days, so i figured what you all thought of it?

Does it seem feasable?

Shotty · 07-11-2005, 04:02 AM

Speed Limits
Automatically generated according to the speed travelled on by each users of the system. Can change along one street. Not sure how you would cope with people who speed and that.

Street Names and Point of Interest
Entered by users. When a map is downloaded and used by another user they can agree with or modify the name of a POI or a Street Name. You would need to introduce something of the concept of 'trust'. The more accurate someone is at entering 'data' the more their entries 'count'.

This would weed out people screwing with the system.

Matthew230 · 07-11-2005, 04:05 AM

I don't think that would work somehow. The way companies do it is to physically drive up and down every road on the map, then sit down and start programming in street names, direction of travel etc. You have also got to have a method of identifying where one road stops and another begins.

xBrady · 07-11-2005, 04:08 AM

Quote: Originally Posted by Matthew230

I don't think that would work somehow. The way companies do it is to physically drive up and down every road on the map, then sit down and start programming in street names, direction of travel etc. You have also got to have a method of identifying where one road stops and another begins.

This would seem to me also the hardest part of creating your own map database. Street names would be very difficult to do.

DeltaFX · 07-11-2005, 04:52 AM

I imagine a small app to scan paper maps, stack your (vectorized) gps log on top of it. Then you can rescale it to match the map. Coz you have a real map to works with, you know where streets end, and you have their name. Once done, that portion of data is made available over p2p. Each day, your puter get new datas, and rebuild itself its database.

A rainy sunday and nothing to do ?, well scan your city map...

Nomad · 07-11-2005, 07:10 AM

That sounds intersting but when you can buy a porgram like MS Street and Trip for about $ 25.00 for the full USA. It would cost more than that for the gas to do a small town.

guge · 07-11-2005, 07:17 AM

I have very often thought of the same thing. I live in Norway, and digital maps are very expensive here. I wrote my own GPS software for use in geocaching and I started logging and plotting GPS data on my laptop when driving. It should be possible to gather and organize data such as these to the point where it becomes a useable roadmap. It would be nice to have a public domain roadmap.

Here is one thing to think about: Such a ploy can only succeed if each user of the system can use it on their own for quite some time. If I'm going to use that software to produce maps from my logs, it almost has to give me instant gratification. Everytime I enter the name of a street, that street should "work" in my local map data.

Uploading and combining would probably be secondary for some time. I don't expect such an initiative to catch on for some time, here in Oslo atleast.

So if someone is already starting to code, concentrate on the user-near components, the server and combining bits can probably wait.

PURDooM · 07-11-2005, 08:44 AM

For the most part the maps come from somewhere public and are parsed into a proprietary form for each companys map software. I myself have been writing map software (just started a new project which is a VB6 open source clone of iguidance, minus the routing for now) and the US source for the maps is the TIGER/LINE (http://www.census.gov/geo/www/tiger/) data and is avaliable for free.

The only flaws I see with the tiger line data over comercial maps is 1. A lack of distinguishing major roads between minor roads (for example tiger describes "sunfish lake blvd" the same as "132nd st", it only makes distinguishing features on higways and freeways), 2. A lack of exit numbers on freeways, 3. A lack of showing which roads are 1 way roads, and 4. A lack of points of interest that isn't parks or schools.

However, I think this data is also publically avaliable as well in different places (I saw a file containing the US highway system which included blvds; http://www.gis.fhwa.dot.gov/gisData.asp), and I can import a phonebook (http://cal.phonelosers.org/forum/news173.htm) and locate different addresses from a phonebook to make a giant POI database with every business/person in a city.

HOWEVER, all that aside I agree it would be nice to have "smart routing software" that learns where paths are on the roads based on where you drive and the direction/speed you drive in, which would be especially usefull for people that live outside the US/europe, but that sounds like alot of work.

Shotty · 07-11-2005, 09:38 AM

I guess the whole idea is that you wouldn't have to do a whole town. It would rely on a constant stream of new information from lots and lots of users. From this it would generate maps.

You wouldn't need to program in things like directions of travel because the system would have already realised it was (for example) a one-way street because the only data it was gotten is going in that one direction on that street.

Only quirks I can see would be the entering of street names, the fact that would need multiple users and to join the maps in some way (otherwise it would only be good at showing you how to get where you've already been

), and stopping people from screwing with the system.

DeltaFX · 07-11-2005, 01:22 PM

Quote: Originally Posted by Nomad

That sounds intersting but when you can buy a porgram like MS Street and Trip for about $ 25.00 for the full USA. It would cost more than that for the gas to do a small town.

I was not clear: I don't see it as "drive all your city then name the streets" for other ppl, that would be stupid, fuc...ly expensive.

I mean, where I live ppl get once a year in their mail box a little book where stores, professionals are listed, where you can find all informartions you may need about city service's, emergency numbers, who and where are doctors, vets, whats pharmacy is open at night... And a little map of the city (A3 papersheet).

Scan that map, use an app to vectorize it (or a part of it if your too lazy), fit your gps log with that map, and let the app recompute a database.

meddler · 07-11-2005, 06:15 PM

Sound like a fantastic idea in theory. The problem is the reliability and ammount of data to get usable maps. I remember talking to someone about this about 6 months ago. To get coverage of a major city it would take quite a lot of people and time. 90% of my driving is local, so my local area would be quite detailed. Unfortunately the area between myself and the next person would be virtually blank, except for maybe the major roads. I'm not saying that everybody would be the same, but I don't think there would be enough people to make it viable.
To check the reliability of the information, youu would have to have someone verify all the data. That is a huge job. I like deltafx idea of overlaying the local street directory to get the start, finish and names of streets, but it is a big job for someone to enter all the information. I'm sure that after a while the people in this project would want to be reimbursed for their time. After all, it would involve a lot of time. You could also have a mic in the car to call the street names as you drive along for varification.
Don't get me wrong, I would like to see free maps, but I don't think they would end up being free. I can see why we have to pay $600 for the maps and the software. There is a great deal of time and effort involved. If this project were to take off, I would be the first one to put up my hand to help, but just the shear size of the task is enough to scare anyone.

Shotty · 07-12-2005, 09:04 AM

There is nothing fun about attampting a simple problem

And I also thought about the same things you're thinking off. You would need alot of a people for it to work. Or a small number of people over a really long time

You could setup a donation system for example, which would pay for some people to spend a day doing a complex area.

Still lots of ideas to think about.

meddler · 07-12-2005, 09:41 PM

maybe you could use something like oziexplorer which you can scan in any type of map and have it mix the info with the vector based data from the people driving around.

kevlar · 07-13-2005, 02:14 AM

In case you weren't aware, a project with Open Source flair was started over a year ago called the Open Street Map. It needs a lot of work, but they're essentially just capturing GPS data, then annotating the data. It's certainly very useful, although I'm sure it'll take a long time until the data is actually useful, attractive, and accurate.

http://www.openstreetmap.org/

I've seen these folks on a mailing list I'm on.

A better solution is really to modify TIGER/2004 data to reflect the Z-layering, one way streets, and up to date information, then formatting the data to be shown in 2D/3D fashion.

I have some talents in these areas, and I would be willing to help, but I have so many projects on my plate that I certainly wouldn't be able to start it myself. Unfortunately, I don't make enough money these days to donate all my time for free.

Kevin

DjDrewDigital · 07-16-2005, 03:31 PM

Hey All,

Have you heard of OxiExplorer? OziExplorer lets you import any type of map. You can scan maps into the system or download them from wherever and import them. You can even import satellite photos or custom maps of an OHV park for example.

If some chick draw's a map to her place on a napkin, you can scan that in and OziExplorer will tell you where you are on the map even if she can't draw straight lines.

http://www.oziexplorer.com/

If you want free maps of california, check out the California Spatial Information Library:

http://gis.ca.gov/data.epl

Since our taxpayer dollars paid for the maps, they are in the public domain and free to download. I downloaded over 1 gig of map data from this site.

User Name		Remember Me?
Password