Community Generated Map Data - Passive Vs. Active / Teleatlas Vs. Openstreetmaps

[Fiberoptic]

I would also like to add that we have been speaking with OpenStreetMap about a collaboration for two months now and the August’s meet is to discuss our collaboration and possibly even show off some concepts if we can get them done in time.

Openstreetmaps has asked us to collect the data and then submit them in large batches.

mp3Car would like not only Malcolm but all other front end developers to write code that allows users of the front end to voluntarily submit GPS tracks to mp3Car.

All data will be publicly available and mp3Car will create a forum where algorithm development can get started right away.

We are also working on some other projects where mp3car users could get free location based services and data in in exchange for the right to use mp3Car crowdsourced map data.

[chunkyks]

*raises hand*

I would be interested in this. For all the unix users out there enjoying gpsd, a small program to log travelled paths is trivial; I will gladly write such a thing myself [although gpsd may already provide that in some manner]. For that matter, depending on needs and requirements, obdgpslogger already does log gps data to an sqlite database, although it lacks a lot of stuff - things I will gladly add to the logging if that would be legitimately useful.

Quote:

In my opinion we need to get all the raw NEMA data to give future algorithms the best chance of success. Let’s talk about that as part of an architecture plan.

*shudder*. The following make for good reading:
Why GPSes suck, and what to do about it
Towards a Better GPS Protocol

Of course, nothing's stopping you from logging NMEA, but I'm inclined to point out that's not as good an idea as some like to think - also, gpsd will upgrade many receivers from NMEA to their own binary protocol [which is much preferable to NMEA, but would mean NMEA sentences simply wouldn't exist to be logged].

My general feeling is it'd be better to standardise on 7-or-14-tuple TPV. But that's probably a topic for a dedicated forum [does such a thing exist yet?]. Alternatively, a shopping list of things that would be conceivably useful to log are in gps_data_t, declared here: http://svn.berlios.de/svnroot/repos/gpsd/trunk/gps.h


I actually have a soft spot in my heart for OSM - I had the idea for obdgpslogger a long time ago, but it wasn't until I went to an OSM meetup that I was galvanised into writing code; OSM provided me a conceptual answer to the problem of what to do with the data once I'd logged it. The problem being that I still haven't actually written an OSM output module for obdgpslogger, since I couldn't figure out a good, reliable, way to do it.

Perhaps an OSM subforum on these forums would be good, with a specific subforum dedicated to discussing the topics [logging for OSM] presented thus far. I would like a place on these forums to investigate ways of writing my OSM output module. obd2osm has been lurking in my TODO list with some thoughts since obdgpslogger began life.


Holy thread derailment, batman.

Gary (-;

PS Oh, yeah, one other thing... It'd be awesome if mp3car were willing to host an SCM repo for code. How exactly projects are started, or how people gain write access, would be another topic for the same OSM subforum?

[jpinkerton]

Count me in on this project, I don't really do software development but would be more than willing to test this out and contribute in any way I can.

[meddler]

maybe Necrolyte could help out, he was working on a gps tracking system.

Sounds like a good idea. I would get involved even if it was just to collect gps data

[Bugbyte]

Quote: Originally Posted by chunkyks

*raises hand*

I would be interested in this. For all the unix users out there enjoying gpsd, a small program to log travelled paths is trivial; I will gladly write such a thing myself [although gpsd may already provide that in some manner]. For that matter, depending on needs and requirements, obdgpslogger already does log gps data to an sqlite database, although it lacks a lot of stuff - things I will gladly add to the logging if that would be legitimately useful.


*shudder*. The following make for good reading:
Why GPSes suck, and what to do about it
Towards a Better GPS Protocol

Of course, nothing's stopping you from logging NMEA, but I'm inclined to point out that's not as good an idea as some like to think - also, gpsd will upgrade many receivers from NMEA to their own binary protocol [which is much preferable to NMEA, but would mean NMEA sentences simply wouldn't exist to be logged].

My general feeling is it'd be better to standardise on 7-or-14-tuple TPV. But that's probably a topic for a dedicated forum [does such a thing exist yet?]. Alternatively, a shopping list of things that would be conceivably useful to log are in gps_data_t, declared here: http://svn.berlios.de/svnroot/repos/gpsd/trunk/gps.h


I actually have a soft spot in my heart for OSM - I had the idea for obdgpslogger a long time ago, but it wasn't until I went to an OSM meetup that I was galvanised into writing code; OSM provided me a conceptual answer to the problem of what to do with the data once I'd logged it. The problem being that I still haven't actually written an OSM output module for obdgpslogger, since I couldn't figure out a good, reliable, way to do it.

Perhaps an OSM subforum on these forums would be good, with a specific subforum dedicated to discussing the topics [logging for OSM] presented thus far. I would like a place on these forums to investigate ways of writing my OSM output module. obd2osm has been lurking in my TODO list with some thoughts since obdgpslogger began life.


Holy thread derailment, batman.

Gary (-;

PS Oh, yeah, one other thing... It'd be awesome if mp3car were willing to host an SCM repo for code. How exactly projects are started, or how people gain write access, would be another topic for the same OSM subforum?

Does it make sense to integrate the solution into LinuxICE?

I think that whatever the 'solution' to passive mapping is, it ought to be fairly simple to implement on the most popular front ends. Maybe as simple as install and almost forget. Upload based on your mp3car user I.D. and so forth.

I'll see what we can do to get the new forum setup so we can start a thread simply to agree what this project is, how it might work, what sort of help mp3car can give, etc. I know they're pretty excited about being able to help the community do something like this.

[soundman98]

I can't program(or understand code for that matter), but would deffinatly be willing to help out where i can.

[Tidder]

Why are all of fiberoptic's posts text so huge? I know he's the owner... but gawsh.

Subscribed.

[malcom2073]

Quote: Originally Posted by Tidder

Why are all of fiberoptic's posts text so huge? I know he's the owner... but gawsh.

Subscribed.

That's how he rolls. Large and in charge.


I'll definitely be keeping an eye on this

[Fiberoptic]

Thanks for all your support guys. We setup a new server for the project today (see other thread) and I have started to help clarify some of the problems that I think exist and the community can solve.

The goal here is high quality free maps and navigation routing that are updated in real time through passive user input.

[kirk78h]

I've been watching these threads with great interest.

Professionally, I build routing and scheduling algorithms for a very large package delivery service (lets leave it at that...). We have been approached by almost all the software mapping companies to use our vehicles as probes. To date, we have not partnered with any of them. As stated earlier in this thread, they want all the data; but do not want to share any of the results.

Map data can get very, very complicated. It seems like there are a lot of bright people contributing to this thread that have read a lot and have insight on how to approach this. I don't mean to be a downer here, but let me throw out a few things to think about. Maybe people here have already solved these issues; maybe not.

1. The geocoder is going to be a big issue (it always is). Given a street address, intersection, etc.; how do you convert that into a Lat, Long?

2. Collecting GPS data is fine; but that is only a very small part of the map layers needed for routing. Intersection information is just as important. Is there a light? a stop sign? Is there a road barrier preventing left hand turns? What is the speed limit? Usually, map data classifies roads as five different types. Is the road one-way? What about connectivity? Looking at a map, an overpass over a road looks like it intersects it.

And this is the tip of the iceburg.

As I said, I think this is a great project; and I'm looking forward to how it progresses.

By the way, my initial interest in putting a computer in my car was a proof-of-concept kind of mission to have the computing power in the vehicle; not as the devices our drivers carry (very expensive and not very powerful).

Kirk

[malcom2073]

Quote: Originally Posted by kirk78h

We have been approached by almost all the software mapping companies to use our vehicles as probes. To date, we have not partnered with any of them. As stated earlier in this thread, they want all the data; but do not want to share any of the results.

Does this mean that you could be potentially interested in using your vehicles as probes for a creative commons license/GPL project like this?

[kirk78h]

Quote: Originally Posted by malcom2073

Does this mean that you could be potentially interested in using your vehicles as probes for a creative commons license/GPL project like this?

No.

[malcom2073]

Quote: Originally Posted by kirk78h

We have been approached by almost all the software mapping companies to use our vehicles as probes. To date, we have not partnered with any of them. As stated earlier in this thread, they want all the data; but do not want to share any of the results.

Apparently I misunderstood what you meant here, could you please elaborate on your reasoning for mentioning this? I would assume that if the reasoning for not partnering with any of them as you said was due to them not wanting to share results, that there would be interest in a project that was not only willing, but the entire purpose was to share results.

[DarquePervert]

Quote: Originally Posted by malcom2073

Does this mean that you could be potentially interested in using your vehicles as probes for a creative commons license/GPL project like this?

Quote: Originally Posted by kirk78h

No.

But would you (or are you) interested in contributing your experience and expertise in working with routing algorithms to any projects that are starting or already underway using OS map data?

Considering your role at "a major package delivery company", it's pretty obvious that you could be helpful to anyone working on such a project. Just wondering if you want to sit on the sidelines and watch or if you want to pitch in and help create something useful for the masses....
Without divulging any trade secrets, of course.

[kirk78h]

Quote: Originally Posted by DarquePervert

But would you (or are you) interested in contributing your experience and expertise in working with routing algorithms to any projects that are starting or already underway using OS map data?

Considering your role at "a major package delivery company", it's pretty obvious that you could be helpful to anyone working on such a project. Just wondering if you want to sit on the sidelines and watch or if you want to pitch in and help create something useful for the masses....
Without divulging any trade secrets, of course.

Yes, I am willing to help. That was the reasoning behind my first post, and listing some of the stumbling blocks I've seen in the past. As I said, we've looked at taking on something like this on several occasions, and always backed off when we evaluated some of these barriers. A good map engine / geocoder is definitely expensive; but our analysis always showed it was cheaper to buy than build. These things, of course, change over time; and the time may be right.

Kirk