Re: [OSM-talk] routing across open spaces
2010/12/7 Anthony : > If all a router knows about is the perimeter, it shouldn't be cutting > through an area. If it understands areas, and the area is tagged as > routable (implicitly or explicitly), then yeah, it should. sorry for joining quite late in this thread, I wanted to point out that there is also a proposal for defining areas implicitly by an relation (which doesn't require a complete perimeter to be mapped): http://wiki.openstreetmap.org/wiki/Relations/Proposed/Area cheers, Martin ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Mon, Dec 6, 2010 at 7:11 PM, David Murn wrote: > On Mon, 2010-12-06 at 18:21 -0500, Anthony wrote: >> On Mon, Dec 6, 2010 at 4:42 PM, David Murn wrote: >> > However, please understand that >> > most of us use routing software, expecting it not to try and take >> > shortcuts across unmapped areas. >> >> Who said anything about taking shortcuts across *unmapped* areas? How >> in the world would that work? > > I was using 'unmapped area', to mean an area marked as (for example) > park, with no other features (ie, pond, trees, paths, barriers, roads) > mapped. A more accurate term for them would be "mapped areas". ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Mon, 2010-12-06 at 18:21 -0500, Anthony wrote: > On Mon, Dec 6, 2010 at 4:42 PM, David Murn wrote: > > However, please understand that > > most of us use routing software, expecting it not to try and take > > shortcuts across unmapped areas. > > Who said anything about taking shortcuts across *unmapped* areas? How > in the world would that work? I was using 'unmapped area', to mean an area marked as (for example) park, with no other features (ie, pond, trees, paths, barriers, roads) mapped. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Mon, Dec 6, 2010 at 6:15 PM, Anthony wrote: > On Mon, Dec 6, 2010 at 3:56 PM, Dave F. wrote: >> On 05/12/2010 22:07, Anthony wrote: >>> - which, if all they >>> know about is the perimeter, is probably a good thing. >> >> Eh? I thought you said you'd "love it" if it cut directly across an area?? > > No, I didn't. Not in that context, anyway. If all a router knows about is the perimeter, it shouldn't be cutting through an area. If it understands areas, and the area is tagged as routable (implicitly or explicitly), then yeah, it should. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Mon, Dec 6, 2010 at 4:42 PM, David Murn wrote: > However, please understand that > most of us use routing software, expecting it not to try and take > shortcuts across unmapped areas. Who said anything about taking shortcuts across *unmapped* areas? How in the world would that work? ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Mon, Dec 6, 2010 at 3:56 PM, Dave F. wrote: > On 05/12/2010 22:07, Anthony wrote: >> >> On Sun, Dec 5, 2010 at 11:03 AM, Dave F. wrote: >>> >>> As long as there are external ways connecting to the area, a router >>> should >>> be able to find the appropriate entrances& exits by tracking the >>> perimeter. >>> I thought they were already able to do that, but maybe not. >> >> Surely they can - just treat it like any other way. > > > >> However, they >> don't treat leisure=park as a routable feature > > All routers? All areas? My understanding is that routers just ignore the area tags completely. So as far as the router knows, so a closed way marked with highway=residential/area=yes is treated exactly the same as any other way marked highway=residential. In other words, it routes along the perimeter, and not through the area itself So allowing routing around the perimeter of an area marked leisure=park would simply require treating leisure=park the same as, say highway=pedestrian. Not that I think this is a good idea. It probably isn't. But it's certainly possible. >> - which, if all they >> know about is the perimeter, is probably a good thing. > > Eh? I thought you said you'd "love it" if it cut directly across an area?? No, I didn't. > They don't have to *follow* the perimeter just use it to find the best exit > & then join it to the entrance to the area with a straight line. > > Are you certain no routers can do that? Of course not. I'm not even certain I know of all routers that exist. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Mon, 2010-12-06 at 21:57 +, Dave F. wrote: > On 06/12/2010 21:42, David Murn wrote: > > Okay, so are we going around the perimeter of the polygon or are we > > taking a straight line cutting directly across an area? > > I think you've deliberately not taken on board many of the points made > in this thread purely to be an argumentative PITA, so you're on your own. I simply asked which of the two options you were proposing we should use, and why you think the problems that have already been discussed and agreed upon dont apply in your instance. Other than because youre going to write a perfect router that does magical Ive taken a lot of the points onboard, infact, if you bother to read back through the thread, youll find that I address almost every single point individually. A look back through the thread shows 37 emails, and you joined the thread at number 31. You'll also find that not only did I take onboard the points, but we even worked out solutions to make routers be able to route properly through these areas. Im guessing since you posted to the thread a week after it happened, that you must have missed bits of it, and are using personal insults to make up for your inability to read. You suggested that "If there are obstructions, then they should be mapped to make OSM more accurate..". You obviously missed the whole concept that routing around objects is easy when you know whats there, the problem is, what do you do when its not mapped? What if someone only maps the area from aerial imagery and doesnt tag anything else? What assumptions do you make, that the area is traversable? > > > ___ > talk mailing list > talk@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On 06/12/2010 21:42, David Murn wrote: On Mon, 2010-12-06 at 21:18 +, Dave F. wrote: On 06/12/2010 21:06, David Murn wrote: On Mon, 2010-12-06 at 20:56 +, Dave F. wrote: - which, if all they know about is the perimeter, is probably a good thing. Eh? I thought you said you'd "love it" if it cut directly across an area?? They don't have to *follow* the perimeter just use it to find the best exit& then join it to the entrance to the area with a straight line. And as was said during the thread, what happens if theres a lake, a building, a playground, etc in the middle of the straight line? As I said earlier in the thread, use multi-polygons. The router *should* be able to get around it (see earlier in the thread about the maths required to get around corners. Okay, so are we going around the perimeter of the polygon or are we taking a straight line cutting directly across an area? I think you've deliberately not taken on board many of the points made in this thread purely to be an argumentative PITA, so you're on your own. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Mon, 2010-12-06 at 21:18 +, Dave F. wrote: > On 06/12/2010 21:06, David Murn wrote: > > On Mon, 2010-12-06 at 20:56 +, Dave F. wrote: > > > >>>- which, if all they > >>> know about is the perimeter, is probably a good thing. > >> Eh? I thought you said you'd "love it" if it cut directly across an area?? > >> > >> They don't have to *follow* the perimeter just use it to find the best > >> exit& then join it to the entrance to the area with a straight line. > > And as was said during the thread, what happens if theres a lake, a > > building, a playground, etc in the middle of the straight line? > > As I said earlier in the thread, use multi-polygons. The router *should* > be able to get around it (see earlier in the thread about the maths > required to get around corners. Okay, so are we going around the perimeter of the polygon or are we taking a straight line cutting directly across an area? > If it can't do this then it's not really fit for purpose & should be > avoided. Its 'not really fit' for your specific purpose, that doesnt mean you should be telling people to avoid it. Should we avoid all routers that dont take into account hgv and maxheight/maxwidth when routing, because its not fit for purpose of driving a big-rig? If you really want fuzzy routing in an application, feel free to add it, thats the whole point of opensource. However, please understand that most of us use routing software, expecting it not to try and take shortcuts across unmapped areas. The biggest problem is if an area is mapped, but the objects in that area arent. If the objects in the park were marked, including paths, then there would be no need for this discussion in the first place. This discussion came up with regards to routing across a park area that has paths but where no paths are mapped. > > What about if the 'straight line' crosses outside of the area, say for > > example if you had an L-shaped area. > > Have you actually read the whole of this thread? Yes, I did, infact I was one half of the monologue when the thread first started, so not only did I read the whole thread, I wrote half of it. Youre the first person to mention a straight line cutting across the area, since everyone explained the problems with it. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On 06/12/2010 21:06, David Murn wrote: On Mon, 2010-12-06 at 20:56 +, Dave F. wrote: - which, if all they know about is the perimeter, is probably a good thing. Eh? I thought you said you'd "love it" if it cut directly across an area?? They don't have to *follow* the perimeter just use it to find the best exit& then join it to the entrance to the area with a straight line. And as was said during the thread, what happens if theres a lake, a building, a playground, etc in the middle of the straight line? As I said earlier in the thread, use multi-polygons. The router *should* be able to get around it (see earlier in the thread about the maths required to get around corners. Even without it should still be able to avoid blockages. If it can't do this then it's not really fit for purpose & should be avoided. What about if the 'straight line' crosses outside of the area, say for example if you had an L-shaped area. Have you actually read the whole of this thread? Dave F. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Mon, 2010-12-06 at 20:56 +, Dave F. wrote: > > - which, if all they > > know about is the perimeter, is probably a good thing. > > Eh? I thought you said you'd "love it" if it cut directly across an area?? > > They don't have to *follow* the perimeter just use it to find the best > exit & then join it to the entrance to the area with a straight line. And as was said during the thread, what happens if theres a lake, a building, a playground, etc in the middle of the straight line? What about if the 'straight line' crosses outside of the area, say for example if you had an L-shaped area. > Are you certain no routers can do that? I think this is what it boils down to, that some routers may be able to do it, but I suspect most cant/wont. As a general rule, routers route directly from one node to another. along a way and only leave that way at a junction. There is no reason you couldnt make a walking router, which doesnt have the restrictions of having to follow a way, but at the moment this isnt how most (all?) of them work. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On 05/12/2010 22:07, Anthony wrote: On Sun, Dec 5, 2010 at 11:03 AM, Dave F. wrote: As long as there are external ways connecting to the area, a router should be able to find the appropriate entrances& exits by tracking the perimeter. I thought they were already able to do that, but maybe not. Surely they can - just treat it like any other way. However, they don't treat leisure=park as a routable feature All routers? All areas? - which, if all they know about is the perimeter, is probably a good thing. Eh? I thought you said you'd "love it" if it cut directly across an area?? They don't have to *follow* the perimeter just use it to find the best exit & then join it to the entrance to the area with a straight line. Are you certain no routers can do that? Dave F. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Sun, Dec 5, 2010 at 11:03 AM, Dave F. wrote: > As long as there are external ways connecting to the area, a router should > be able to find the appropriate entrances & exits by tracking the perimeter. > I thought they were already able to do that, but maybe not. Surely they can - just treat it like any other way. However, they don't treat leisure=park as a routable feature - which, if all they know about is the perimeter, is probably a good thing. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On 01/12/2010 00:48, David Murn wrote: On Tue, 2010-11-30 at 19:14 -0500, Anthony wrote: That's nonsense. A way does not show a right of passage. A particularly tagged way shows a right of passage. And a park is a particularly tagged way. No, a park *CAN BE* a particularly tagged way. Can be? How can you represent a park in OSM without using a way which is tagged with leisure=park? Okay, a park that you can route through, is a particularly tagged park. The point still remains that with incomplete tagging, you have to make assumptions about whether the park is traversable. Just like a road, if it isnt tagged properly with oneway/access/barrier/etc, the routing will be inaccurate. Correct. What are you getting at? See above Actually, the fact that its not tagged correctly is a big part of the issue. The renderer has to make assumptions if its not tagged. If there was a tagging scheme to indicate that an area was traversable, then routing engines could start to use it, There is such a tagging scheme. It is described at http://wiki.openstreetmap.org/wiki/Access There is also a tagging scheme footway=yes or highway=path. Simply putting an access=yes tag onto an area doesnt give you any more information about it. It doesnt tell you if theres a barrier or gate, it doesnt tell you if theres a big pond or lake smack-bang in the middle of the park which you have to walk around. but Id hate for a routing engine to try and take a short cut 'as the crow flies' through an area which hasnt got ways marked to follow. I'd love it. It's a feature I'm quite looking forward to. If you want routers to route through unmapped areas, then you can simply ignore the directions given and keep following the map, once you emerge out the other side of the park, the routing can carry-on from wherever you happen to come out. One day OSM will be able to route me from Linkwood Avenue to Pine Bay Drive through the park (http://www.openstreetmap.org/?lat=28.07187&lon=-82.550402&zoom=18&layers=M), saving me 50 minutes of walking. Imagine if you tried to save 50min by getting routed across Albert Park[1]. That big thing in the middle of the screen is a lake that extends almost the entire length of the park, hence you'll notice all the walking paths have been mapped in, to allow you to be routed through the park appropriately. The way to avoid getting a soaking is to use multipolygon relations (as has already been done for the lakes island). It creates a rough doughnut shape. An area way such as this park doesn't *need* linear ways to cross it. An area has an infinite number of ways criss-crossing it. If there are obstructions, then they should be mapped to make OSM more accurate.. As long as there are external ways connecting to the area, a router should be able to find the appropriate entrances & exits by tracking the perimeter. I thought they were already able to do that, but maybe not. Cheers Dave F. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
I think you should just use the shape outline to create 2 possible routes, and present the geographic average between the 2. The average may be weighted differently if there are large differences in length between the 2. Comments in the route description will show that an area is to be crossed, and so will the user act when he is actually presented with the area ;<) -Oorspronkelijk bericht- Van: talk-boun...@openstreetmap.org [mailto:talk-boun...@openstreetmap.org] Namens David Murn Verzonden: zondag 5 december 2010 2:36 Aan: Anthony CC: OSM Talk Onderwerp: Re: [OSM-talk] routing across open spaces On Wed, 2010-12-01 at 00:00 -0500, Anthony wrote: > Anyway, I looked around at a few places labelled leisure=park, and the > usage is all over the place. I'd say based on that very unscientific > sample that it's probably best for routers to use a default of > access=unknown for leisure=park areas, and only use parks for short > cuts if they're explicitly tagged with something like foot=permissive. > > Alternatively, I guess it wouldn't be horrible to add something like a > highway=shortcut tag, so mappers could be explicit about it. If we've > gotta add foot=permissive by hand anyway, it's not that much more work > to add a few extra ways. I was thinking of this issue last night while playing with my routing software. One issue I thought about which makes this difficult, is different shapes. If youre trying to route across a square area or any area which has a direct path from start to finish, routing in your method is easy. But I was thinking, what happens if youve got an L-shaped park or even a U-shaped park. +-++---+ | || | | || +-C-+ | A | | BA | | | | B +-++-+ +-+ Do you cross the open part of the area in the second example, in a straight line from A to B? Do you form an arc? Do you simply go from A to B, but go around the edge near C? As an area is unlikely to be a perfect oblong, this situation may arise more often than one would think. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Wed, 2010-12-01 at 00:00 -0500, Anthony wrote: > Anyway, I looked around at a few places labelled leisure=park, and the > usage is all over the place. I'd say based on that very unscientific > sample that it's probably best for routers to use a default of > access=unknown for leisure=park areas, and only use parks for short > cuts if they're explicitly tagged with something like foot=permissive. > > Alternatively, I guess it wouldn't be horrible to add something like a > highway=shortcut tag, so mappers could be explicit about it. If we've > gotta add foot=permissive by hand anyway, it's not that much more work > to add a few extra ways. I was thinking of this issue last night while playing with my routing software. One issue I thought about which makes this difficult, is different shapes. If youre trying to route across a square area or any area which has a direct path from start to finish, routing in your method is easy. But I was thinking, what happens if youve got an L-shaped park or even a U-shaped park. +-++---+ | || | | || +-C-+ | A | | BA | | | | B +-++-+ +-+ Do you cross the open part of the area in the second example, in a straight line from A to B? Do you form an arc? Do you simply go from A to B, but go around the edge near C? As an area is unlikely to be a perfect oblong, this situation may arise more often than one would think. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Wed, 2010-12-01 at 00:00 -0500, Anthony wrote: > Alternatively, I guess it wouldn't be horrible to add something like a > highway=shortcut tag, so mappers could be explicit about it. If we've > gotta add foot=permissive by hand anyway, it's not that much more work > to add a few extra ways. I think you may be onto something there. If you added a highway or footway tag to the area, to explicitly show routability, then routers may be more able to use it reliably. Another thought that comes to mind is that routing normally occurs from node to node, so if a triangle area was formed with 3 nodes, a routing engine couldnt generate a route that crosses through the middle, as a routing engine will only route from one node to another on the same way, or that shares a junction. You could get around this by adding footways into the park area, and then convince the routing engine authors to recognize a highway (or footway) tag on an area. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
>> >> One day OSM will be able to route me from Linkwood Avenue to Pine Bay >> >> Drive >> >> through the park >> >> (http://www.openstreetmap.org/?lat=28.07187&lon=-82.550402&zoom=18&layers=M), >> >> saving me 50 minutes of walking. >> > >> > Imagine if you tried to save 50min by getting routed across Albert >> > Park[1]. That big thing in the middle of the screen is a lake that >> > extends almost the entire length of the park, hence you'll notice all >> > the walking paths have been mapped in, to allow you to be routed through >> > the park appropriately. >> >> You forgot the link, so I'm going to imagine that the lake was mapped. >> Obviously a lake is by default foot=no. > > My mistake, sorry. The link I tried to post before, is > http://www.openstreetmap.org/?lat=-37.84597&lon=144.97075&zoom=16 > > If you look at aerial imagery, you'll see most of the park is open > grassland and the lake, however there is a path network in the park, > which is appropriately mapped and tagged. Right. Clearly the lake is not routeable, and any routing engine would have to route around the lake :). The fact that the non-routability of the lake takes precedence over the routability of the park would be for the same reason that a lake is rendered on top of a park. Anyway, I looked around at a few places labelled leisure=park, and the usage is all over the place. I'd say based on that very unscientific sample that it's probably best for routers to use a default of access=unknown for leisure=park areas, and only use parks for short cuts if they're explicitly tagged with something like foot=permissive. Alternatively, I guess it wouldn't be horrible to add something like a highway=shortcut tag, so mappers could be explicit about it. If we've gotta add foot=permissive by hand anyway, it's not that much more work to add a few extra ways. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, 2010-11-30 at 20:24 -0500, Anthony wrote: > >> One day OSM will be able to route me from Linkwood Avenue to Pine Bay Drive > >> through the park > >> (http://www.openstreetmap.org/?lat=28.07187&lon=-82.550402&zoom=18&layers=M), > >> saving me 50 minutes of walking. > > > > Imagine if you tried to save 50min by getting routed across Albert > > Park[1]. That big thing in the middle of the screen is a lake that > > extends almost the entire length of the park, hence you'll notice all > > the walking paths have been mapped in, to allow you to be routed through > > the park appropriately. > > You forgot the link, so I'm going to imagine that the lake was mapped. > Obviously a lake is by default foot=no. My mistake, sorry. The link I tried to post before, is http://www.openstreetmap.org/?lat=-37.84597&lon=144.97075&zoom=16 If you look at aerial imagery, you'll see most of the park is open grassland and the lake, however there is a path network in the park, which is appropriately mapped and tagged. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, Nov 30, 2010 at 7:48 PM, David Murn wrote: > On Tue, 2010-11-30 at 19:14 -0500, Anthony wrote: > >> >> > That's nonsense. A way does not show a right of passage. A >> >> > particularly tagged way shows a right of passage. And a park is a >> >> > particularly tagged way. >> > >> > No, a park *CAN BE* a particularly tagged way. >> >> Can be? How can you represent a park in OSM without using a way which >> is tagged with leisure=park? > > Okay, a park that you can route through, is a particularly tagged park. A way which is tagged with leisure=park is a particularly tagged way. > The point still remains that with incomplete tagging, you have to make > assumptions about whether the park is traversable. Correct. The same is true of a way tagged with highway=footway or highway=steps or highway=proposed or highway=motorway, if it doesn't have any access tags, of course. > See above Indeed. >> > Actually, the fact that its not tagged correctly is a big part of the >> > issue. The renderer has to make assumptions if its not tagged. If >> > there was a tagging scheme to indicate that an area was traversable, >> > then routing engines could start to use it, >> >> There is such a tagging scheme. It is described at >> http://wiki.openstreetmap.org/wiki/Access > > There is also a tagging scheme footway=yes or highway=path. Simply > putting an access=yes tag onto an area doesnt give you any more > information about it. Sure it does, it tells you the area is a right of way. > It doesnt tell you if theres a barrier or gate, > it doesnt tell you if theres a big pond or lake smack-bang in the middle > of the park which you have to walk around. Well, no, of course not. Neither does footway=yes, or highway=path, access=yes, tell you if there's a barrier or gate or a big pond or lake in the middle of the footway/path. If someone maps an area, sticks access=yes on it, and doesn't map the gates, the routers can't be expected to work correctly. Whether or not access=permissive (not access=yes) should be the default is something I can't really say, because I haven't taken a sample to see how common it is for parks with fences to mapped as leisure=park and with no mention of the fence. >> > but Id hate for a routing >> > engine to try and take a short cut 'as the crow flies' through an area >> > which hasnt got ways marked to follow. >> >> I'd love it. It's a feature I'm quite looking forward to. > > If you want routers to route through unmapped areas, then you can simply > ignore the directions given and keep following the map, once you emerge > out the other side of the park, the routing can carry-on from wherever > you happen to come out. I don't want them to route through unmapped areas. I want them to route (via foot) through mapped areas tagged with leisure=park. >> One day OSM will be able to route me from Linkwood Avenue to Pine Bay Drive >> through the park >> (http://www.openstreetmap.org/?lat=28.07187&lon=-82.550402&zoom=18&layers=M), >> saving me 50 minutes of walking. > > Imagine if you tried to save 50min by getting routed across Albert > Park[1]. That big thing in the middle of the screen is a lake that > extends almost the entire length of the park, hence you'll notice all > the walking paths have been mapped in, to allow you to be routed through > the park appropriately. You forgot the link, so I'm going to imagine that the lake was mapped. Obviously a lake is by default foot=no. >> what if someone marked a national park as an area, should routers simply >> show a route ATCF or should it only route through a national park area >> if there is a way to follow? >> >> Routers shouldn't route through all areas the same, any more than they >> should route along all non-areas the same. An area tagged with >> leisure=park wouldn't have the same access defaults as an area tagged >> with building=yes. > > Okay, so what areas are routable then? Is leisure=playground passable? Not sure what the default for that should be. Probably foot=permissive. > What about parking areas, Default is probably foot=permissive, vehicle=destination. > golf courses, schools, graveyards, farms? I'd say the default there is access=private. > All > of these places can sometimes be travelled across, but not all. These > places often also have paths to follow, as do most parks, especially > between entrances. You don't seem to be using the term "park" the same way as is intended by leisure=park. leisure=park is meant for open areas. Just because something is called a park doesn't mean it should be tagged with leisure=park. In any case, as you allude to, routing of vehicles through a parking lot generally shouldn't use shortest path. For foot traffic, shortest path is probably a good approximation. Thanks for the discussion. You raise a lot of good points about how much detail will have to go into the routers in order to properly handle this. I have no doubt one day that detail will be added, tho
Re: [OSM-talk] routing across open spaces
On Tue, 2010-11-30 at 19:14 -0500, Anthony wrote: > >> > That's nonsense. A way does not show a right of passage. A > >> > particularly tagged way shows a right of passage. And a park is a > >> > particularly tagged way. > > > > No, a park *CAN BE* a particularly tagged way. > > Can be? How can you represent a park in OSM without using a way which > is tagged with leisure=park? Okay, a park that you can route through, is a particularly tagged park. The point still remains that with incomplete tagging, you have to make assumptions about whether the park is traversable. > > Just like a road, if it > > isnt tagged properly with oneway/access/barrier/etc, the routing will be > > inaccurate. > > Correct. What are you getting at? See above > > Actually, the fact that its not tagged correctly is a big part of the > > issue. The renderer has to make assumptions if its not tagged. If > > there was a tagging scheme to indicate that an area was traversable, > > then routing engines could start to use it, > > There is such a tagging scheme. It is described at > http://wiki.openstreetmap.org/wiki/Access There is also a tagging scheme footway=yes or highway=path. Simply putting an access=yes tag onto an area doesnt give you any more information about it. It doesnt tell you if theres a barrier or gate, it doesnt tell you if theres a big pond or lake smack-bang in the middle of the park which you have to walk around. > > but Id hate for a routing > > engine to try and take a short cut 'as the crow flies' through an area > > which hasnt got ways marked to follow. > > I'd love it. It's a feature I'm quite looking forward to. If you want routers to route through unmapped areas, then you can simply ignore the directions given and keep following the map, once you emerge out the other side of the park, the routing can carry-on from wherever you happen to come out. > One day OSM will be able to route me from Linkwood Avenue to Pine Bay Drive > through the park > (http://www.openstreetmap.org/?lat=28.07187&lon=-82.550402&zoom=18&layers=M), > saving me 50 minutes of walking. Imagine if you tried to save 50min by getting routed across Albert Park[1]. That big thing in the middle of the screen is a lake that extends almost the entire length of the park, hence you'll notice all the walking paths have been mapped in, to allow you to be routed through the park appropriately. > what if someone marked a national park as an area, should routers simply > show a route ATCF or should it only route through a national park area > if there is a way to follow? > > Routers shouldn't route through all areas the same, any more than they > should route along all non-areas the same. An area tagged with > leisure=park wouldn't have the same access defaults as an area tagged > with building=yes. Okay, so what areas are routable then? Is leisure=playground passable? What about parking areas, golf courses, schools, graveyards, farms? All of these places can sometimes be travelled across, but not all. These places often also have paths to follow, as do most parks, especially between entrances. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, Nov 30, 2010 at 7:18 PM, Robin Paulson wrote: > On 1 December 2010 13:14, Anthony wrote: >> I'd love it. It's a feature I'm quite looking forward to. One day >> OSM will be able to route me from Linkwood Avenue to Pine Bay Drive >> through the park >> (http://www.openstreetmap.org/?lat=28.07187&lon=-82.550402&zoom=18&layers=M), >> saving me 50 minutes of walking. > > you realise you can walk that route without OSM's blessing? No, I didn't realize that. However, now that I do, I'll still love it when a routing program helps me find similar shortcuts that I don't already know about. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On 1 December 2010 13:14, Anthony wrote: > I'd love it. It's a feature I'm quite looking forward to. One day > OSM will be able to route me from Linkwood Avenue to Pine Bay Drive > through the park > (http://www.openstreetmap.org/?lat=28.07187&lon=-82.550402&zoom=18&layers=M), > saving me 50 minutes of walking. you realise you can walk that route without OSM's blessing? -- robin http://tangleball.co.nz/ - Auckland's Creative Space http://bumblepuppy.org/blog/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, Nov 30, 2010 at 6:57 PM, David Murn wrote: > On Wed, 2010-12-01 at 12:01 +1300, Robin Paulson wrote: > >> > That's nonsense. A way does not show a right of passage. A >> > particularly tagged way shows a right of passage. And a park is a >> > particularly tagged way. > > No, a park *CAN BE* a particularly tagged way. Can be? How can you represent a park in OSM without using a way which is tagged with leisure=park? > Just like a road, if it > isnt tagged properly with oneway/access/barrier/etc, the routing will be > inaccurate. Correct. What are you getting at? > Actually, the fact that its not tagged correctly is a big part of the > issue. The renderer has to make assumptions if its not tagged. If > there was a tagging scheme to indicate that an area was traversable, > then routing engines could start to use it, There is such a tagging scheme. It is described at http://wiki.openstreetmap.org/wiki/Access > but Id hate for a routing > engine to try and take a short cut 'as the crow flies' through an area > which hasnt got ways marked to follow. I'd love it. It's a feature I'm quite looking forward to. One day OSM will be able to route me from Linkwood Avenue to Pine Bay Drive through the park (http://www.openstreetmap.org/?lat=28.07187&lon=-82.550402&zoom=18&layers=M), saving me 50 minutes of walking. > To take that to a further level, > what if someone marked a national park as an area, should routers simply > show a route ATCF or should it only route through a national park area > if there is a way to follow? Routers shouldn't route through all areas the same, any more than they should route along all non-areas the same. An area tagged with leisure=park wouldn't have the same access defaults as an area tagged with building=yes. And an area representing a national park should not be tagged with leisure=park. This is explicitly stated at http://wiki.openstreetmap.org/wiki/Park "Parks in rural locations (e.g. areas named "National parks") are a totally different kind of thing. leisure=nature_reserve and boundary=national_park should be used, and not this leisure=park tag." ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Wed, 2010-12-01 at 12:01 +1300, Robin Paulson wrote: > > That's nonsense. A way does not show a right of passage. A > > particularly tagged way shows a right of passage. And a park is a > > particularly tagged way. No, a park *CAN BE* a particularly tagged way. Just like a road, if it isnt tagged properly with oneway/access/barrier/etc, the routing will be inaccurate. > this is all true guys, but it's getting away form the point: parks are > only one implementation/manifestation of the situation. i'm enquiring > about routing across areas in general, and whether anyone does/will do > it. One example of this, could be parking areas. Shopping centres for example have organised parking areas, with defined ways and spaces. Other areas, for example a showground, might simply have a big dirt parking area. Or a playing field, which in suburban areas might be an open field that you can walk across, but for a more formal field or stadium, youre unlikely to be able to walk across. > whether or not this is across a park is only a small (and possibly > irrelevant) part of the issue. whether or not the area has been tagged > correctly is only part of the issue Actually, the fact that its not tagged correctly is a big part of the issue. The renderer has to make assumptions if its not tagged. If there was a tagging scheme to indicate that an area was traversable, then routing engines could start to use it, but Id hate for a routing engine to try and take a short cut 'as the crow flies' through an area which hasnt got ways marked to follow. To take that to a further level, what if someone marked a national park as an area, should routers simply show a route ATCF or should it only route through a national park area if there is a way to follow? David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, Nov 30, 2010 at 6:08 PM, Anthony wrote: > Just take the n ways which connect to the area, and > run > http://www.loria.fr/~lazard/Publications/Curvature-constrained_shortest_path_in_a_convex_polygon/Curvature-constrained_shortest_path_in_a_convex_polygon.html Actually, that was a more difficult problem. This simpler one would also work: http://alienryderflex.com/shortest_path/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, Nov 30, 2010 at 6:01 PM, Robin Paulson wrote: > but it's getting away form the point: parks are > only one implementation/manifestation of the situation. i'm enquiring > about routing across areas in general, and whether anyone does/will do > it. I don't think anyone currently does do it. I expect that eventually someone will. So I'd recommend tagging now in preparation for that eventuality. Whether or not access=permissive or access=no is the default is not even that important. If you're not sure which will be the standard, just be redundant. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, Nov 30, 2010 at 1:24 PM, Felix Hartmann wrote: > I don't think routing over areas will ever work well. It might work for 2-3 > areas and that is already really difficult for calculation. Just routing > onto an open space is no problem, but calculating a route over a longer > distance, would mean that for any place the time it needs and the way to > take have to be calculated as a way, before using it for autorouting. Hence > in the end the only thing that will work, is that there are invisible ways > for routing with very low priority added to the map, because everything else > won't be feasible for longer distances. Why low priority? Just take the n ways which connect to the area, and run http://www.loria.fr/~lazard/Publications/Curvature-constrained_shortest_path_in_a_convex_polygon/Curvature-constrained_shortest_path_in_a_convex_polygon.html > As long as there is no consensus or widespread use, of how map rendering > should add such invisible ways, no area should be considered to be > routable. How can there ever be widespread use if no areas are considered routable? Sounds like a chicken and egg problem to me. > Real polygon routing is AFAIK with the currently developed algos impossible > (but also does not exist in reality. I don't understand what you mean by "real polygon routing". We're talking about a router here, so the only route that matters is the shortest path through the area which takes you from one way connected to an area to another way connected to an area. Given n such ways, there are n choose 2 such paths, which can be calculated via the algorithm I linked to above. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On 1 December 2010 11:55, Anthony wrote: > "A park. Open, green area for recreation, usually municipal. These are > outdoor areas, typically grassy/green areas, set aside of leisure and > recreation. > Typically (or pretty much always) open to the public, but may be > fenced off, and may be closed e.g. at night time." > >> This is the problem. A way on the map, shows there is a right of >> passage. If theres no way in an area, it might mean the area is >> passable, but it might also mean it hasnt been mapped. Hence, the >> safest option is to not route unless theres a right of way to follow. > > That's nonsense. A way does not show a right of passage. A > particularly tagged way shows a right of passage. And a park is a > particularly tagged way. this is all true guys, but it's getting away form the point: parks are only one implementation/manifestation of the situation. i'm enquiring about routing across areas in general, and whether anyone does/will do it. whether or not this is across a park is only a small (and possibly irrelevant) part of the issue. whether or not the area has been tagged correctly is only part of the issue -- robin http://tangleball.co.nz/ - Auckland's Creative Space http://bumblepuppy.org/blog/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, Nov 30, 2010 at 5:22 PM, David Murn wrote: > On Tue, 2010-11-30 at 11:43 -0500, Anthony wrote: >> On Tue, Nov 30, 2010 at 11:29 AM, David Murn wrote: >> > On Tue, 2010-11-30 at 15:30 +, Ed Avis wrote: >> >> As a rough rule, leisure=park and landuse=grass could be considered >> >> walkable, >> >> unless tagged access=no or access=private. >> > >> > You may also find timed access restrictions apply. Some parks here for >> > example, are gated off at night, which again may affect vehicle/bicycle >> > traffic but not foot traffic, or it may affect everything. >> >> Fortunately, there are tags for all of this: >> >> http://wiki.openstreetmap.org/wiki/Access#Transport_mode_restrictions >> http://wiki.openstreetmap.org/wiki/Access#Access_time_restrictions >> http://wiki.openstreetmap.org/wiki/Barrier > > Thats great, what happens if someone traces the park from aerial > imagery, and doesnt know/care about any of those? Or for that matter, > if someone tags it on-the-ground, but doesnt add all the details? Then the mapping will be incomplete until someone else comes along and fixes it. Same as what happens if, for instance, someone traces a street from aerial imagery and doesn't know/care about the one-way tag, or the "no motorcycles" restriction. > What defautls should routers use? Should they assume all parks are flat > and walkable, or should they assume theyre fenced, or that theyre 'stay > off the grass' or 'stay on the grass' or what? Personally, I'd assume that they're (relatively) flat and walkable. >From the wiki: "A park. Open, green area for recreation, usually municipal. These are outdoor areas, typically grassy/green areas, set aside of leisure and recreation. Typically (or pretty much always) open to the public, but may be fenced off, and may be closed e.g. at night time." > This is the problem. A way on the map, shows there is a right of > passage. If theres no way in an area, it might mean the area is > passable, but it might also mean it hasnt been mapped. Hence, the > safest option is to not route unless theres a right of way to follow. That's nonsense. A way does not show a right of passage. A particularly tagged way shows a right of passage. And a park is a particularly tagged way. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On 1 December 2010 11:22, David Murn wrote: > Thats great, what happens if someone traces the park from aerial > imagery, and doesnt know/care about any of those? Or for that matter, > if someone tags it on-the-ground, but doesnt add all the details? > > What defautls should routers use? Should they assume all parks are flat > and walkable, or should they assume theyre fenced, or that theyre 'stay > off the grass' or 'stay on the grass' or what? > > This is the problem. A way on the map, shows there is a right of > passage. If theres no way in an area, it might mean the area is > passable, but it might also mean it hasnt been mapped. Hence, the > safest option is to not route unless theres a right of way to follow. your arguments are all true, but they're no different to any other situation with incomplete data. and there is no 'should' for routing engines. each engine designer chooses his/her own method for solving the problem. that's how free/open works -- robin http://tangleball.co.nz/ - Auckland's Creative Space http://bumblepuppy.org/blog/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, 2010-11-30 at 11:43 -0500, Anthony wrote: > On Tue, Nov 30, 2010 at 11:29 AM, David Murn wrote: > > On Tue, 2010-11-30 at 15:30 +, Ed Avis wrote: > >> As a rough rule, leisure=park and landuse=grass could be considered > >> walkable, > >> unless tagged access=no or access=private. > > > > You may also find timed access restrictions apply. Some parks here for > > example, are gated off at night, which again may affect vehicle/bicycle > > traffic but not foot traffic, or it may affect everything. > > Fortunately, there are tags for all of this: > > http://wiki.openstreetmap.org/wiki/Access#Transport_mode_restrictions > http://wiki.openstreetmap.org/wiki/Access#Access_time_restrictions > http://wiki.openstreetmap.org/wiki/Barrier Thats great, what happens if someone traces the park from aerial imagery, and doesnt know/care about any of those? Or for that matter, if someone tags it on-the-ground, but doesnt add all the details? What defautls should routers use? Should they assume all parks are flat and walkable, or should they assume theyre fenced, or that theyre 'stay off the grass' or 'stay on the grass' or what? This is the problem. A way on the map, shows there is a right of passage. If theres no way in an area, it might mean the area is passable, but it might also mean it hasnt been mapped. Hence, the safest option is to not route unless theres a right of way to follow. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On 30.11.2010 17:17, Anthony wrote: On Tue, Nov 30, 2010 at 10:53 AM, Robert Kaiser wrote: Robin Paulson schrieb: or am i missing a tag? do i need to tag parks, etc. with "area=yes" "foot=yes", "access=yes" or would that be a case of "tagging for the routing engine" Note that in some park, stepping on the grass is explicitely forbidden, so automatically routing across park space may pose a problem. What is the default, though. For a leisure=park, area=yes is certainly the default. access=yes certainly is not (you typically can't drive through a park). As for foot=*, I'd say foot=permissive is the default (from the definition - "Typically (or pretty much always) open to the public, but may be fenced off, and may be closed e.g. at night time."). If stepping on the grass is explicitly forbidden, the area should be marked with foot=no or foot=private. As for the ability of routers to utilize open spaces for routing, I think that's more of a future feature than a current one, regardless of how it is tagged. Doesn't seem like it would be all that difficult, as a preprocessor would just need to add implicit ways connecting the possible routes through the park. But in any case, doesn't really matter much how it's tagged, so long as non-routable park areas are tagged as such. I wouldn't apply the same logic to wooded areas, of course, but those wouldn't be properly tagged with leisure=park. As for long rows of bushes, those should be tagged - maybe barrier=hedge? ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk I don't think routing over areas will ever work well. It might work for 2-3 areas and that is already really difficult for calculation. Just routing onto an open space is no problem, but calculating a route over a longer distance, would mean that for any place the time it needs and the way to take have to be calculated as a way, before using it for autorouting. Hence in the end the only thing that will work, is that there are invisible ways for routing with very low priority added to the map, because everything else won't be feasible for longer distances. As long as there is no consensus or widespread use, of how map rendering should add such invisible ways, no area should be considered to be routable. Real polygon routing is AFAIK with the currently developed algos impossible (but also does not exist in reality. If you put a camera over a public place, and record 24hours, you'll soon notice even though anyone is free to walk wherever he wants to, actually there are more or less defined pathes that everyone takes when crossing a place - hence one could also argue, that such undefined but existing pathes should be mapped, but with a consensus that they are not visbibly rendered in maps, and also not shown in editors by default, except when say activating an editing session with the goal of working strictly on autorouting features -- which is in my eyes how such areas should be delt with ultimately (though not yet, because the editors still show everything and a myriad of pathes over places would make editing much harder)) ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, Nov 30, 2010 at 6:14 PM, Martijn van Exel wrote: > On Tue, Nov 30, 2010 at 4:50 PM, Nic Roets wrote: > >> On Tue, Nov 30, 2010 at 2:22 AM, Robin Paulson >> wrote: >> >>> hi, >>> i walk a lot, and would like a routing engine which understands i can >>> take a direct route across an open public space, such as a park, >>> without needing a footpath to be explicitly drawn in. the existing >>> routing engines don't seem to understand this. >>> >>> or am i missing a tag? do i need to tag parks, etc. with "area=yes" >>> "foot=yes", "access=yes" or would that be a case of "tagging for the >>> routing engine" >>> >>> >> Firstly note that routing across areas is (theoretically) much harder than >> routing along ways (Non-polynomial time VS polynomial time). >> >> Secondly note that the problem is not restricted to pedestrian routing, >> e.g. parking areas. There have been cases where people mapped the road >> surface as areas, although they would then also have ways running down the >> centerline. >> >> Supporting areas is on my list of things that I would like to do, but >> there are many other things in front of it. I recently added dragable routes >> to the Osm.org Routing Demo. I improved the endpoints. Negotiated for a >> better server. Routing instructions and translations. And for Christmas I >> want a mobile application for large scale collection of house numbers. >> >> >> I definitely second your call for a mobile app to easily collect house > numbers. A single-purpose app could be very simple. Would you want to have > something graphical (user pinpoints address on map on-screen) or something > even simpler (user enters housenumber, selects street based on location or > accepts best match, address node is sent to OSM). In the latter case, which > is what I would prefer, how would you deal with GPS inaccuracy? > > Non-graphical. The streetname is not added to the address PoI. Nominatim is smart enough to take the closest street. If I travel down a street it, I want to be able to tell it that house 23 is on my left using just 3 or 4 keystrokes / taps. http://help.openstreetmap.org/questions/1385/what-is-the-best-mobile-application-for-large-scale-house-number-collection > Martijn van Exel +++ m...@rtijn.org > laziness – impatience – hubris > http://schaaltreinen.nl | http://martijnvanexel.nl | > http://oegeo.wordpress.com/ > twitter / skype: mvexel > flickr: rhodes > > > ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
David Murn incanberra.com.au> writes: >>Not all park land is walkable - some can be trees or bushes - so some extra >>tagging is needed. > >Another problem, is that you may not be able to traverse the park in all >directions. It may have a fence with only a couple of access gates, or >you may find that some access gates are suitable for wheeled vehicles >where other gates may have steps or other obstacles, so you may find >yourself being routed to a corner of the park which you cant easily get >out of. Yes, and at present there is an implicit understanding that only the explicitly mapped paths are routable - at least where I live. In London the city parks have entrance footways added from the street, and where there is no entrance that usually means there isn't a gate to go in, so by omission there must be a fence or wall. But it would be better to map the fence or wall explicitly. If the routing engines start to route across grass and park land then it might prompt mappers to add the missing features. >You may also find timed access restrictions apply. Yes, but those can be handled using the usual tags, just as there may be timed access restrictions for a highway or gate. -- Ed Avis ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, Nov 30, 2010 at 11:29 AM, David Murn wrote: > On Tue, 2010-11-30 at 15:30 +, Ed Avis wrote: >> As a rough rule, leisure=park and landuse=grass could be considered walkable, >> unless tagged access=no or access=private. > > You may also find timed access restrictions apply. Some parks here for > example, are gated off at night, which again may affect vehicle/bicycle > traffic but not foot traffic, or it may affect everything. Fortunately, there are tags for all of this: http://wiki.openstreetmap.org/wiki/Access#Transport_mode_restrictions http://wiki.openstreetmap.org/wiki/Access#Access_time_restrictions http://wiki.openstreetmap.org/wiki/Barrier ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, 2010-11-30 at 15:30 +, Ed Avis wrote: > Not all park land is walkable - some can be trees or bushes - so some extra > tagging is needed. Another problem, is that you may not be able to traverse the park in all directions. It may have a fence with only a couple of access gates, or you may find that some access gates are suitable for wheeled vehicles where other gates may have steps or other obstacles, so you may find yourself being routed to a corner of the park which you cant easily get out of. > As a rough rule, leisure=park and landuse=grass could be considered walkable, > unless tagged access=no or access=private. You may also find timed access restrictions apply. Some parks here for example, are gated off at night, which again may affect vehicle/bicycle traffic but not foot traffic, or it may affect everything. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, Nov 30, 2010 at 10:53 AM, Robert Kaiser wrote: > Robin Paulson schrieb: >> >> or am i missing a tag? do i need to tag parks, etc. with "area=yes" >> "foot=yes", "access=yes" or would that be a case of "tagging for the >> routing engine" > > Note that in some park, stepping on the grass is explicitely forbidden, so > automatically routing across park space may pose a problem. What is the default, though. For a leisure=park, area=yes is certainly the default. access=yes certainly is not (you typically can't drive through a park). As for foot=*, I'd say foot=permissive is the default (from the definition - "Typically (or pretty much always) open to the public, but may be fenced off, and may be closed e.g. at night time."). If stepping on the grass is explicitly forbidden, the area should be marked with foot=no or foot=private. As for the ability of routers to utilize open spaces for routing, I think that's more of a future feature than a current one, regardless of how it is tagged. Doesn't seem like it would be all that difficult, as a preprocessor would just need to add implicit ways connecting the possible routes through the park. But in any case, doesn't really matter much how it's tagged, so long as non-routable park areas are tagged as such. I wouldn't apply the same logic to wooded areas, of course, but those wouldn't be properly tagged with leisure=park. As for long rows of bushes, those should be tagged - maybe barrier=hedge? ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, Nov 30, 2010 at 4:50 PM, Nic Roets wrote: > On Tue, Nov 30, 2010 at 2:22 AM, Robin Paulson wrote: > >> hi, >> i walk a lot, and would like a routing engine which understands i can >> take a direct route across an open public space, such as a park, >> without needing a footpath to be explicitly drawn in. the existing >> routing engines don't seem to understand this. >> >> or am i missing a tag? do i need to tag parks, etc. with "area=yes" >> "foot=yes", "access=yes" or would that be a case of "tagging for the >> routing engine" >> >> > Firstly note that routing across areas is (theoretically) much harder than > routing along ways (Non-polynomial time VS polynomial time). > > Secondly note that the problem is not restricted to pedestrian routing, > e.g. parking areas. There have been cases where people mapped the road > surface as areas, although they would then also have ways running down the > centerline. > > Supporting areas is on my list of things that I would like to do, but there > are many other things in front of it. I recently added dragable routes to > the Osm.org Routing Demo. I improved the endpoints. Negotiated for a better > server. Routing instructions and translations. And for Christmas I want a > mobile application for large scale collection of house numbers. > > > I definitely second your call for a mobile app to easily collect house numbers. A single-purpose app could be very simple. Would you want to have something graphical (user pinpoints address on map on-screen) or something even simpler (user enters housenumber, selects street based on location or accepts best match, address node is sent to OSM). In the latter case, which is what I would prefer, how would you deal with GPS inaccuracy? Martijn van Exel +++ m...@rtijn.org laziness – impatience – hubris http://schaaltreinen.nl | http://martijnvanexel.nl | http://oegeo.wordpress.com/ twitter / skype: mvexel flickr: rhodes ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
Robin Paulson schrieb: or am i missing a tag? do i need to tag parks, etc. with "area=yes" "foot=yes", "access=yes" or would that be a case of "tagging for the routing engine" Note that in some park, stepping on the grass is explicitely forbidden, so automatically routing across park space may pose a problem. Robert Kaiser ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
On Tue, Nov 30, 2010 at 2:22 AM, Robin Paulson wrote: > hi, > i walk a lot, and would like a routing engine which understands i can > take a direct route across an open public space, such as a park, > without needing a footpath to be explicitly drawn in. the existing > routing engines don't seem to understand this. > > or am i missing a tag? do i need to tag parks, etc. with "area=yes" > "foot=yes", "access=yes" or would that be a case of "tagging for the > routing engine" > > Firstly note that routing across areas is (theoretically) much harder than routing along ways (Non-polynomial time VS polynomial time). Secondly note that the problem is not restricted to pedestrian routing, e.g. parking areas. There have been cases where people mapped the road surface as areas, although they would then also have ways running down the centerline. Supporting areas is on my list of things that I would like to do, but there are many other things in front of it. I recently added dragable routes to the Osm.org Routing Demo. I improved the endpoints. Negotiated for a better server. Routing instructions and translations. And for Christmas I want a mobile application for large scale collection of house numbers. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] routing across open spaces
Robin Paulson gmail.com> writes: >I walk a lot, and would like a routing engine which understands i can >take a direct route across an open public space, such as a park, >without needing a footpath to be explicitly drawn in. the existing >routing engines don't seem to understand this. > >or am i missing a tag? do i need to tag parks, etc. with "area=yes" >"foot=yes", "access=yes" or would that be a case of "tagging for the >routing engine" There are pedestrian areas in cities (rendered by Mapnik as a kind of grey blob) which I assume routing engines can manage. But by convention parks in OSM have explicit footways marked across them, and I guess routing engines rely on that rather than allowing you to walk anywhere. Not all park land is walkable - some can be trees or bushes - so some extra tagging is needed. But I think this unwalkable land is the exception so it's that land that should be tagged, rather than adding foot=yes areas for almost everywhere. I suppose a routing engine would impose a small penalty for walking directly across grass, so explicit paths would still be favoured if they exist. I suggest you file bugs against the routing engines and see what they say. As a rough rule, leisure=park and landuse=grass could be considered walkable, unless tagged access=no or access=private. -- Ed Avis ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
[OSM-talk] routing across open spaces
hi, i walk a lot, and would like a routing engine which understands i can take a direct route across an open public space, such as a park, without needing a footpath to be explicitly drawn in. the existing routing engines don't seem to understand this. or am i missing a tag? do i need to tag parks, etc. with "area=yes" "foot=yes", "access=yes" or would that be a case of "tagging for the routing engine" cheers -- robin http://tangleball.co.nz/ http://bumblepuppy.org/blog/ ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk