|
In several places that I have mapped recently, a road either leads into a car park or forms part of the boundary of a parking area, and a footpath starts from a different part of the same car park. Does that make the road and the footpath effectively joined, and if not, what is the recommended way to map them? A similar question to this has been asked before (Car parks footpaths and auto-routing), but it did not seem to result in a definitive answer. |
|
It depends a little on the exact layout. I assume you are trying to indicate that it is possible to travel along the road, through the car park, and then onto the footpath (e.g. for cyclists). One option is to mark a service road in the car park (highway=service, service=parking_aisle), and join the footpath to the service road. Here is an example of this (although with cycleway rather than footpath): http://osm.org/go/uN96oklGS-- This looks like the best approach. Where the car park is really an unstructured parking area, no great articificality would be introduced by bringing all paths and access roads to one or more notional meeting points. For a more organised car park, perhaps with marked lanes, we have to hope that its designers provided for pedestrians, so that there are some genuine footpaths that can be mapped. I will go ahead with this approch unless anyone objects.
(26 Jan '12, 20:01)
Madryn
2
Connecting the paths to the service roads (where they are connected) is good. Connecting them arbitrarily is bad. But not as bad as missing roads/paths. If you insist on connecting them somewhere within the parking lot, put there at least some fixme/note saying that you made the roads up to make routing work. There is a popular misleading saying about how one should not tag, I never liked it. My advice:
(26 Jan '12, 23:29)
LM_1
|
|
You could test it yourself if you have route and map capable device, or give us a location and somebody will.To link a location do this, right click permalink and copy and paste the link address into a comment. 1
I tried using MapSource to find a pedestrian route from the road entrance to Thatchers Copse car park (http://www.openstreetmap.org/?lat=50.83063&lon=-1.25046&zoom=17&layers=M) to the footpath at its south west corner. It routed round the North side of Thatchers Copse, ignoring the possibility of walking through the car park.
(26 Jan '12, 19:02)
Madryn
If you're using MapSource, then you're using a Garmin map, and so if you were to create your own map based on OSM data it should be possible to enforce pretty much any routability you want. Are you downloading the Garmin map from somewhere or creating it yourself using mkgmap or something else?
(26 Jan '12, 19:19)
SomeoneElse ♦
I took the map from Talkytoaster's web site. I'm still something of an OSM novice, and I haven't yet worked out how to make my own maps directly from OSM data. I'm not concerned about pedestrian routability just for myself; rather, I would like anything that I add to OSM to be right for as many people as possible. Making my own maps with routability that suits me therefore does not seem to be a full solution to the problem. At present I like Ebenezer's suggestion, i.e. join roads and paths through unstructured areas if necessary.
(26 Jan '12, 21:17)
Madryn
1
I notice that both paths are unconnected to the car park you could try connecting them, but even when connected it will be a while before Talkytoaster's excellent work is updated,every two weeks? before you could then upload the new map and test the results.
(27 Jan '12, 10:37)
andy mackey
My apologies for missing that rather obvious point. An answer in a few weeks is better than no answer at all, so I have connected both footpaths to the boundary of the car park. Does the OSM database have the concept of 'open access' areas, where walkers (or in the case of rural car parking areas, drivers) are not required to keep to specific paths? Does digital mapping in general have that concept? In fairness, it took England several hundred years to come up with the idea!
(06 Feb '12, 19:39)
Madryn
http://help.openstreetmap.org/questions/2985/access-land-tag you could ask again in a new question or comment on mine there may be some new info
(06 Feb '12, 20:00)
andy mackey
Thanks to the continuing good work of Talkytoaster, I have re-tested the routing on MapSource, and as expected it won't create a pedestrian route across a car park even when all paths are joined to the car park's perimeter. That leaves two possibilities: 1 - Extend the access road into the car park, and connect the footpaths to it. 2 - leave the paths unconnected. I favour option 1, as its only error is that it shows a single track where any number of lines are possible. Option 2 seems worse because it causes routing software to make incorrect decisions.
(14 Feb '12, 21:09)
Madryn
1
... or option 3, fix the router to work with the map rather than "tagging for the router". In this case this might even be possible via a mkgmap style file.
(15 Feb '12, 00:26)
SomeoneElse ♦
If it is possible to fix the router, it is beyond my ability. I have noticed the problem on MapSource and on an Etrex 30. Also, consider some of the replies to a related question (http://help.openstreetmap.org/questions/3177/car-parks-footpaths-and-auto-routing), such as 'the bad news is that I have never heard of any production quality street map routing software or firmware that can route through a polygon' and 'I'd consider every router that uses parking aisles as shortcuts to be quite broken'.
(15 Feb '12, 18:44)
Madryn
showing 5 of 9
show 4 more comments
|
