[Openstreetmap] New roads appearing without GPS tracks
simon at zymurgy.org
Fri Dec 16 10:21:07 GMT 2005
Graham Wall wrote:
> I am pleased to see that the status table in the wiki has spurred
> progress with the motorways. Last night a large section of the M4
> through Newport and Cardiff was filled in, both carriageways, with
> There appear to be very few GPS data points along certain sections and
> junctions and was wondering if these have been deleted immediately
> after the section was completed, as I myself have done in the past.
> Take a look at this section near Cardiff:
> Recently I have taken to leaving all my data points on the server, for
> future reference, but also as a means of accountability. It shows not
> only that the roads have a non-copyrighted source, but also gives
> confidence; if someone else edits the same section of road, they will
> have a reference to why I put it there in the first place.
> Another possibility is that the data were entered using an offline
> editor. This again raises the questions of accountability and
> confidence. If the source data are not on the central server, proving
> data were not copied from copyrighted maps becomes more difficult.
I've been editing some of the data near Newport, and I have been working
on the premise that:
I *personally know* that that the M4 now has two carriageways all the
way to Pont Abraham, and that beyond that, there's dual carriageway on
the A48 to Carmarthen, and dual carriageway on the A40 to the roundabout
at St Clears.
I *personally know* the road, having driven it many times (without
carrying a GPS)
I *personally know* the approximate locations of the junctions, and the
shape of many of the junctions.
From the landsat images (which we *can* use as reference), you can
frequently make out the route of the roads, though the accuracy may not
be as good as if there were GPS tracks.
Yes, it would be far better if every single slip road was driven by GPS
carrying drivers, and if those GPS logs were of a better sampling rate
and/or the actual track lines between track points were drawn), but I
don't think we've got enough GPS wielding volunteers in some parts of
I've also been thinking that for doing things like motorway junctions,
it helps to have a picture in your mind what the junction looks like,
and in some cases, I've used timelapse video footage (cheap webcam and
laptop style) of someone (eg myself) driving the route in order to
recall that there really is a junction at a particular point, even if I
didn't actually drive down that exit.
Suggestions as to how to record such surveying in the database (without
uploading the entire video) are welcome!
More information about the talk