Thread: Crowdsourced Maps Project - mods help/thoughts

Originally Posted by Bugbyte

Well, I'm not sure that leaving mapping and routing algorithms to professional companies because it is hard or time consuming or both is a very good argument for the main issue:

It wouldn't be-if anyone was making that argument. Your not competing against a corporation, or re-inventing the wheel for open source, your blatantly ignoring all the work of one very good, very open and very free database to start things from scratch. If you could do it better that might be one thing, but you guys don't even have anyone that can do waypoint calculation.

Originally Posted by Bugbyte

Map data and routing engines remain proprietary. Adding tracks to OpenStreet maps isn't automatic. Passive mapmaking is possible, navigation corporations are already doing it.

Again-Open Street Maps and Navit - completely open source (aka not proprietary). Your absolutely right, and ten minutes on an automatic track submission app (something that could be integrated into plugins or run in the background) would be a great addition to OSM's arsenal.

Originally Posted by Bugbyte

This project seeks to make it easier to collect tracks and add them to maps. That's step one.

And it was a good step....step 2 is where you started to head the wrong way.....and step three is where you dove off the cliff.

Originally Posted by Bugbyte

Step two can process those tracks and add them to OpenStreetmap if we want, or use OpenStreetMap to supplement the tracks. Or both if someone wants to. Point is, they're free to use.

So mp3car has no problem providing the 20GB, 40GB , 60GB etc data dumps to other projects as the collection grows? Unless you guys have unlimited bandwidth and a few T3's thats gonna be very expensive.

Originally Posted by Bugbyte

People have spent thousands of hours reinventing the wheel to make things open source, from operating systems to car pc interfaces. A community project to help make nav data free is precisely the type of project that could easily attract outside interests and help push the whole thing forward.

And it did - OSM has over 170,000 users. They started with people that knew what they were doing, they were the first to make this data open and free and they now have maps good enough for everyday use.

You guys are attempting to re-invent the wheel just because you can, not due to any need or lack of open source alternatives. Not sure how many open source projects you guys have worked on, but I can tell you from experience where the ones that try figuring things out as they go along end up.

justchat: How are we re-inventing the wheel? There isn't any open source software out there which can take automatically collected tracks and turn them into OSM map data. There are proprietary programs to do this, but nothing open source. I thought the purpose of this was to help OSM out by creating OSM ready maps without all the extra effort and hassle? Did I misunderstand?

Originally Posted by malcom2073

We're trying to contribute, not to parallel.

Originally Posted by Heather

In terms of reinventing the wheel, why not try generating free maps and routing engines here on the forums if people want to.

Like I said...You need to figure out the goal first.

And no offense to the programmers on the forum but hacking together a plugin to a front end and designing a full navigation engine are not both in the "casual side projects" category.

justchat: How are we re-inventing the wheel? There isn't any open source software out there which can take automatically collected tracks and turn them into OSM map data. There are proprietary programs to do this, but nothing open source. I thought the purpose of this was to help OSM out by creating OSM ready maps without all the extra effort and hassle? Did I misunderstand?

That comment was directed at the suggestions that the forum could start its own map database and maybe create its own nav software.

I fully support the idea of creating an automated submission system for Open Street Maps. It would be a tool, a small app you make in an hour and release open source to everyone. Yes it would still require people to check the tracks and make any map changes but thats a fact of how these things work.

The idea of taking a series of points in space and turning them into a map isn't just far fetched its not possible. No matter how automated this system is, someone still has to identify what streets are what, using any source to do this automatically would be stealing from that source and would most likely result in a lawyer up your ***.

Originally Posted by lambosprit

But I have to say that until you start, this kind of thing never gets moving, which is why I've started.

Using sequential numbers like you've been doing is going to limit any error correction the system could do - your better off using a sequential timestamp.

Ok in that case, I agree. Trying to create our own navigation software or even a simple routing engine is years away. The purpose of this project afaik was to automate the process of collecting and contributing map data. It may not be 100% automated, but it could reduce the workload to the point where it would take one person where before it would take hundreds. If someone wants to put together a project to re-create navit (yes, it does need to be recreated before it becomes really useful) that's fine but I don't think that conversation is even remotely related to this project. Sure that could be part of a larger project, but trying to focus on too broad an objective as justchat said is a surefire way to fail.

As for data collection, the goal is irrelevant. I've said this a multitude of times before, but it must have gotten lost. The focus should be collecting ALL gps information, and then eliminating what we don't need at a later point as we develop programs and algorithms for sorting through the data. It's easier to delete data later than to recover what we never had, or waste months figuring out what we need ahead of time, only to find out months later we were wrong. There's no such thing as too much data.

Originally Posted by malcom2073

As for data collection, the goal is irrelevant. I've said this a multitude of times before, but it must have gotten lost. The focus should be collecting ALL gps information, and then eliminating what we don't need at a later point as we develop programs and algorithms for sorting through the data. It's easier to delete data later than to recover what we never had, or waste months figuring out what we need ahead of time, only to find out months later we were wrong. There's no such thing as too much data.

Well yes and no. If we stored every string outputted by a gps receiver mp3car would never have the bandwidth to handle all the uploaded data. So instead we parse some of it, for some reasons its information that will never be useful (like satellite health),for other cases its for privacy, like the date/time.

There is also the huge issue of what data is returned....you don't wan't data submitted when GPS signal isn't up to par because that harms the integrity of the whole database. At the same time things like HDOP and VDOP could play a critical part in snapping to a road but only if they fall in acceptable limits.

And most important of all...server side storage should not be in xml text files! I keep saying that again and again but nothing changed.

Originally Posted by malcom2073

Ok in that case, I agree. Trying to create our own navigation software or even a simple routing engine is years away. The purpose of this project afaik was to automate the process of collecting and contributing map data. It may not be 100% automated, but it could reduce the workload to the point where it would take one person where before it would take hundreds.

Let's figure out how to automagically contribute to tracks for OpenStreetMaps. Doing that allows us to both contribute to the OSM project but also, if we are so inclined, to have tracks that we can do whatever we want with (terrain profiling comes to mind).

How about this? Let's do two things simultaneously:

1. Start collecting data by testing Lambosprit's server. The volume of data isn't a problem right now because NOBODY IS UPLOADING IT. We need some data and some testers.

2. Start a thread on the best way to store the data. We already had a discussion on this, but let's hash out what the right storage schema should be. We can always convert the existing data to that format before we get too much.

agreed.

there we go...now everyones on the same page.

As far as data storage and interoperability - whos in contact with the OSM developers? Can we find out if .GPX files are easiest for them or if they would prefer batch adding via an API? Also, can we batch add under a single username (ex: mp3car) or will they want unique usernames?

Fiberoptic talked with them. I'll ping him and have him ask about it.