Real-Time Collaborative Mapmaking

savetz writes "Throughout October and November, citizens of Amsterdam were invited to wear a GPS tracker as they went about their lives. Their whereabouts were forwarded to a server which created a map of the city in real-time, based on "the sheer movements of real people." The site includes aggregate maps, and those created by individuals, including a subway driver, cyclist, and marathon trainer."

Comments

GPSr fun

[faeryman]

I was sitting around yesterday thinking about what else can be done with GPS besides geocaching, etc. This is pretty cool.

If you like this, you might like GPS drawing too.

Excellent!

[Marxist Commentary]

I've been using a Garmin GPS II+ that I originally bought for mapping backcountry nonmotorized trails, using a mountain bike. It was great for that, including its ability to upload its waypoints and track log to my PC, where I could integrate the data into mapmaking (GIS) software.

On the water, I found its articulated antenna and lack of waterproofing a liability. I used it in a housing (heavy-duty ziploc type), but found
it awkward. I just got a new Garmin eTrex, a 12-channel waterproof (1m for 30 min) unit that has the features I want. No internal maps, but I've never seen a handheld GPS that can compete with
the big-screen units in that department-- they're as good as an actual NOS chart.

If it were possible to do so, I would love to donate my map data to such a project - perhaps an open source map project? It could make for some exciting orienteering!

Re:Excellent!

[flippet]

That would certainly be useful for those times when you're hopelessly lost, find yourself and spend ages wondering exactly where you went.

I spent two hours last month hacking round in a forest after going off the track while cycling; I'd _love_ to know where I went, and how close to various paths I must have gone...

Phil, just me

Reminds me of Cryptonomicon

[devphil]

There's a great scene in Cryptonomicon where he writes about attaching lights to people's heads, then tracking the lights from a side view as they walk on and off street curbs. Graphing over time would produce a square wave, with sidewalks showing up as high lines, and streets as low lines.

The point is that Waterhouse is one of the kinds of people who could stare at boatloads of those graphs, and then emerge with an extremely detailed street map of London.

Re:reminds me of the half a car joke

[nairnr]

This would be the case if only cars went passed an particular intersection. In real life, you would have both the sensor, and then a study where people actually do a car count to determine a reasonable ratio for car / truck traffic. Think about it. The sensor is only measuring axle passages. Cars may have two, but trucks can have anywhere from 3 axles up.

If you ever wonder why there are a number of cars sitting on the grass around intersections or interchanges that is what they are doing.

It would be a lot more fun to sit at the sensor and go back and forward...

I've been thinking about something similar

[Randy Rathbun]

One thing I have noticed about a lot of open source stuff is that they all use those horrific Tiger maps. Tiger maps suck. Big time. They are good for just looking at street names, but when you want to use them for anything more than that they are not good at all. For example, the street I live on shows up on the tiger maps as being about 50 feet south of where it really is, and the intersection closest to my house is more than 150 feet off.

But, because tiger maps are free, that is what is used. However, if you want to build your own talking GPS navigation system, using those maps is a bad thing - the routes would be all goofed up.

So, I have been considering doing something like these guys so there could be some way to get decent maps out there for free that will give better results.

However, I see a number of potential problems. First, you can't just turn on a GPS and start recording data. My Magellan 315 starts averaging when you sit still. Get under an obstruction and the data it records is all over the place. After only an hour of sitting you would have somewhere around 100 data points that don't exist. That would be a lot of work for someone to clean up.

The other problem I see is really related to participation. It's one thing if you are in a small town with 50 or so streets in it - you would probably be more apt to get your town fully mapped. However, living in the Big City, I only really travel a few streets during a typical month - mainly highways and interstates, with the occassional side street. Since the ways I travel always have a ton of traffic on them, that means most everyone else goes that way too - probably including a good number of people who might be participating in such a project as well.

Re:I've been thinking about something similar

[SEWilco]

So ignore the data points which are close together during a long period. You're not particularly interested in data from an hour when you're not moving. Throw away data from periods when you're not moving or when it has been "too long" since a GPS reading.

If the data is coming from a bunch of people who are wandering around for weeks, they'll probably run through the same area several times and generate several valid data points. Even more if they keep doing it for months.

Matching the raw data to actual locations is a different problem. Let's say we start with TIGER maps. Use speeds to locate and confirm the highway locations. People who enter/exit a freeway will show where the limited number of access points are, so now the location of those intersecting streets is known (if TIGER doesn't show the ramps, you might have to record them -- easy to do that if you care). Cars which use those known streets allow correlation of TIGER locations of those streets. Turns and cross traffic show intersecting streets -- counting number of blocks from highways allows confirmation of location of all individual streets (alleys would add some confusion). Streets with curves in them allow further confirmation of locations along the street. Patterns of traffic pauses show STOP signs or traffic lights. Areas with only interpolated data hint at tree cover or tall buildings.

Remote controlled eyes

[jki]

...this story reminds me of one the dark-hour ideas. Would not be sometimes handy if you could "rent the eyes" of people around the world, for say 5 minutes. Just like in the GPS'ing Amsterdarm scenario, there would be people in multiple cities with world - carrying GPS and a digital (video) camera.

Then, when you spot someone in a interesting place, you could get control of his camera for some time. Maybe there could be the option donate the guy $5 and ask him to move 100 feet left.... now that's a real big brother scenario :) Anyway, it could be fun too, to be "in the place" when something interesting happens...talking about Amsterdam...(red) lights, camera, action :)

Two-timers need (should) not apply

[Black Copter Control]

I can just see it now, Your spouse looks up your GPS data and finds that just about every night you said you were 'working late', you have thes long traces at their best friend's house

(Sorry: Former best friend).

subterrestrial GPS?

[flowerp]

So how exactly does a subway driver get a GPS
signal on the job?

And how do Australians get GPS? They're 6000 miles
below our feet!

I did this last year

[jjeffers]

I did something like this last year using amateur radio's Automatic Packet Reporting system. Basically there are thousands of amateur radio operators that broadcast their GPS position. Using software I wrote along with MySQL and Mapserver I was able to create maps that showed real-life traffic flow.

You can see a sample of this type of map and learn a little bit about it at http://aprsworld.net/info/paper2002/giant-map-of-l a.php

-James Jefferson
KB0THN

Find me

Accuracy, precision and QA of spatial data

[xixax]

One thing I have noticed about a lot of open source stuff is that they all use those horrific Tiger maps. Tiger maps suck. ... and the intersection closest to my house is more than 150 feet off.

That's one of the cans of worms I have been pondering deeply ever since I demo-ed a PostGIS based system that lets random users lodge spatial data.How on earth do you do meaningful QA on that quantity of data? Especially when many people are happy to use layers such as TIGER.

FWIW, I think that collecting raw GPS tracks is a low value activity. I think the most value will come from validated overlays put together in a Wiki fashion. Leave it to the Cave Clanners to maintain a Wiki style layer of drains and urban tunnels, leave it to the geeks to maintain a WiFi layer. Pull you favorite overlay sites together in your PDA and enjoy an augmented reality...

Xix.

Beyond Open Source, what about Open Data?

[Travelr9]

I think it's an excellent idea to create some sort of structured repository for geophysical data. To draw an analogy, I love Google, but it worries me that Google owns their (increasingly universal) dataset as a private for-profit corporation. Map data about the globe is just about the most 'open' sort of information that a human could need... getting from here to there, and information about here or there, being pretty fundamental to life. There are all sorts of applications -- travel, leisure, commerce, play, whatever.

The OSS community does a great job thinking about the pros and cons of open source *software* but doesn't do nearly so much work on the subject of open *data* with which to use the software. This issue is implicit in Slashdot favorites such as FOIA requests to reign in government, public domain vs. copyright issues, fair use, etc., but no one seems to formally frame the 'Open Data' issue in a way that motivates the same sort of response as does Open Source. We should start a new movement, parallel and complementary to OSS! Any takers? We need bandwidth and storage, just like everyone else... I see from Internic that mapdata.org is available. Someone want to hack out a good MySQL table structure to hold it all? Lemme know when we open the project on SourceForge... etc.

Be active!

fps levels anyone?

[haa...jesus christ]

am i the only person that thinks we could generate fps levels using this technique? yes? well okay then.