Re: [OSM-talk] Clarifying and representing road markings at junctions
Picking up Ray's point that observing the back of the giveway sign is a rather indirect way of saying follow the road round to the right, the simplest/clearest is probably a relation on the through route linking the ways before/after the junction, saying this is the priority route through the junction. Maybe simply a type=priority relation, with no roles? I'd probably use this as well as marking the giveways on the junction arms (giveway=forward/backward on a node seems to express it succinctly; who knows if it'll catch on). But mebbe I should also file a Potlatch trac ticket that allows you to paste a single tag/relation from memory. I don't need to be pasting a set of 20 relations on a way that already has 19 of them. Richard ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Clarifying and representing road markings at junctions
On 30/09/2009 22:05, Roy Wallace wrote: On Wed, Sep 30, 2009 at 7:35 PM, David Earl da...@frankieandshadow.com wrote: On 30/09/2009 10:20, Martin Koppenhoefer wrote: you could model it like this (see attached, colours are just indicating the ways, not highway-classes) Yes, that's also what I typically do, e.g. http://www.openstreetmap.org/?lat=52.596517lon=0.376144zoom=18layers=B000FTF Eek. Nice hack, but dodgy... 1) What if the road name changes *at* the junction, not just after the junction? 2) That hack just seems to change two things: a) it changes the *angle* between the intersecting ways at the junction. Is there any reason to want to do this? What exact problem does it solve? It's not a hack, it's a very reasonable representation of what's on the ground. The kerb line may be straight on one side but is usually curved opposite the junction, and many such junctions now have a build out which reinforces the curve around the corner, but even if it isn't the centre line curves around the corner, and often the give way lines as well. It shows visually which the main road is at the junction and is a good model of the physical arrangement. b) it makes a single *way* continue through the intersection. err, no. If the road has the same name around the corner it can do. If the minor but straight on road has the same name it could be a continuous way, but I would normally break it at that point, not least so the name of the minor road is clear. But where ways break is of no significance - you have to break ways at all sorts of places because of changes in the environment like speed limits, starts of bridges etc. Does this actually infer that there is no giveway instruction? Not necessarily, though there nearly always is. If so, is this documented anywhere? (I'm sure I could find examples where this is not the case) If not, then the hack *doesn't* explicitly show that the curved road continues through the intersection without interruption. I think anyone looking at it would understand the arrangement on the ground, and it does model the situation as I see it. It is very unusual indeed in the UK anyway to find a case where priority is around the corner but there is no curvature at all in the way it goes around the corner. If there really is no curve whatsoever (and I can't think of an example off hand that I've mapped in 3 years of mapping, though I'm sure there are some), then I wouldn't try to model a curve. It's not a hack, it's what's actually there to a greater or lesser extent in most such circumstances. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Clarifying and representing road markings at junctions
On Thu, Oct 1, 2009 at 8:05 PM, David Earl da...@frankieandshadow.com wrote: It shows visually which the main road is at the junction and is a good model of the physical arrangement. IMHO it does not *explicitly* show the continuations of roads at the junction. And even if you do think it works visually, that is not sufficient - there are many uses for OSM data. b) it makes a single *way* continue through the intersection. err, no. If the road has the same name around the corner it can do. If the minor but straight on road has the same name it could be a continuous way, but I would normally break it at that point, not least so the name of the minor road is clear. But where ways break is of no significance - you have to break ways at all sorts of places because of changes in the environment like speed limits, starts of bridges etc. Ah, so are you saying that, in Martin's attached image, the red way and the yellow way should/could meet at the junction? If so, then IMHO it is even *less* clear that, e.g. traveling from the red to the grey way is a left turn, whereas traveling from the red way to the yellow way is uninterrupted. I think anyone looking at it would understand the arrangement on the ground, and it does model the situation as I see it. Don't be fooled - people are not the only ones that look at OSM data. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Clarifying and representing road markings at junctions
On 01/10/2009 11:47, Roy Wallace wrote: On Thu, Oct 1, 2009 at 8:05 PM, David Earl da...@frankieandshadow.com wrote: It shows visually which the main road is at the junction and is a good model of the physical arrangement. IMHO it does not *explicitly* show the continuations of roads at the junction. And even if you do think it works visually, that is not sufficient - there are many uses for OSM data. b) it makes a single *way* continue through the intersection. err, no. If the road has the same name around the corner it can do. If the minor but straight on road has the same name it could be a continuous way, but I would normally break it at that point, not least so the name of the minor road is clear. But where ways break is of no significance - you have to break ways at all sorts of places because of changes in the environment like speed limits, starts of bridges etc. Ah, so are you saying that, in Martin's attached image, the red way and the yellow way should/could meet at the junction? If so, then IMHO it is even *less* clear that, e.g. traveling from the red to the grey way is a left turn, whereas traveling from the red way to the yellow way is uninterrupted. No I was referring to the real examples I quoted. I think anyone looking at it would understand the arrangement on the ground, and it does model the situation as I see it. Don't be fooled - people are not the only ones that look at OSM data. I don't understand this at all. I am just mapping what I see on the ground. And please don't patronise, I'm well aware of the uses of OSM data and have contributed to many of them. The main road goes round a corner (and may or may not share the same name). I represent the corner even though there may be a straight kerb line on one side, when curvature exists e.g. on the opposite kerb or in the white lining. The slight curve before the side road branches off might possibly allow a bright routing algorithm to describe it more accurately. But I think there would be ambiguity here independent of any mapping, because straight on is somewhat ambiguous, especially when it's not a complete right angle turn as sometimes happens. In English I think I'd want to be told follow the road round to the left or some such in these circumstances, not a simple turn left. A T junction certainly wouldn't achieve that possibility without more information. An explicit tagging would. But in the absence of that, modelling what's on the ground goes some way. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Clarifying and representing road markings at junctions
On Thu, Oct 1, 2009 at 9:35 PM, David Earl da...@frankieandshadow.com wrote: Ah, so are you saying that, in Martin's attached image, the red way and the yellow way should/could meet at the junction? If so, then IMHO it is even *less* clear that, e.g. traveling from the red to the grey way is a left turn, whereas traveling from the red way to the yellow way is uninterrupted. No I was referring to the real examples I quoted. A picture would help my brain...sorry. I think anyone looking at it would understand the arrangement on the ground, and it does model the situation as I see it. Don't be fooled - people are not the only ones that look at OSM data. I don't understand this at all. I am just mapping what I see on the ground. And please don't patronise, I'm well aware of the uses of OSM data and have contributed to many of them. Apologies if that came across patronising - I really didn't mean it that way. My point is that IMHO mapping so that it is understood when looked at by a person is not sufficient (as, it seems, you're already well aware). The main road goes round a corner (and may or may not share the same name). I represent the corner even though there may be a straight kerb line on one side, when curvature exists e.g. on the opposite kerb or in the white lining. That's all fine. I'm just saying that *doesn't* indicate that following the curved road *doesn't* constitute a turn. In English I think I'd want to be told follow the road round to the left or some such in these circumstances, not a simple turn left. Exactly - for software to be able to say, e.g. follow the road around to the left/right, the mapper needs to be able to map continuations of roads at junctions. This can be done explicitly, using the methods (relations) I described. Using a curve does not do this explicitly. A T junction certainly wouldn't achieve that possibility without more information. An explicit tagging would. But in the absence of that, modelling what's on the ground goes some way. We agree here. Modeling what's on the ground (e.g. curves, etc.) goes some way, but does not answer the original poster's question. For that, we seem to agree that we need explicit tagging. What did you think of my suggestions? ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Clarifying and representing road markings at junctions
you could model it like this (see attached, colours are just indicating the ways, not highway-classes) cheers, Martin attachment: Bildschirmfoto.png___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Clarifying and representing road markings at junctions
You could record it as a type of turn-restriction relation, but I have a prejudice against those, having copied them down a bus route for quite a way until I realised I'd picked up a stray. That (of course) may be a problem with the editor I'm using, but keeping it simple is always a good maxim. So I'd much prefer a giveway instruction (giveway=yes or giveway=-1) on the way that gives way, probably on a node near the junction, and inferring the direction from the way that the node is on. I lost the will to read when much the same issue was discussed with regard to stop signs, so I don't know what the conclusion was (if any). Richard ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Clarifying and representing road markings at junctions
On 30/09/2009 10:20, Martin Koppenhoefer wrote: you could model it like this (see attached, colours are just indicating the ways, not highway-classes) Yes, that's also what I typically do, e.g. http://www.openstreetmap.org/?lat=52.596517lon=0.376144zoom=18layers=B000FTF Even though the kerb doesn't have a bend in it, if the road markings indicate the priority goes round the corner I generally put a kink in the way to show this. Though that example, the road name carries round the corner too, that's not always the case. Here, for example: http://www.openstreetmap.org/?lat=52.330664lon=0.344934zoom=18layers=B000FTF Brook Street is also the name of the little stub, but the road markings make it clear that the priority is around the corner into Tanners Lane. David ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Clarifying and representing road markings at junctions
On Tuesday 29 September 2009 21:59:10 Matt Williams wrote: I've been noticing recently a problem we're going to/already have in our data when it comes to routing directions particularly. It concerns how to define continuations of roads at junctions and/or the road markings that delineate that. This problem manifests itself in many ways but for a first example, look at the attached image (road.png). On the left you will see the physical plan of a road junction near where I live. The way that it would be represented in the OSM data model is shown on the right. In this case, it would be sensible to make a way out of the segments 'a' and 'b' (yeah, I know we don't have segments any more, it's just an explanation tool), call it, e.g. 'Curve Road' and make a second way out of segment 'c' and call it 'Small Road'. At this stage, the date representation is sound and routing application would have no problem knowing how to parse it. However, there are two (increasingly common) ways in which this model will be forced to be broken: I don't see where problem lies. Is it that routing software will not be able to choose right route? You never stated it clearly, but if I understand correctly road from segment a to b has right of way over segment c. So all you need is a way to indicate this. There are some proposals that could solve this (stop signs, yield, right of way). If there are turn restrictions you can map those using turn restriction relation. But if there is no explicit right of way and no turn restrictions it really should not matter how road markings are painted on the road. Routing software should be able to pick the right route based on other criteria (road classification, speed limits, traffic calming, ...). Of course if roads b and c lead to completely different destinations (they don't join for several kilometers) it should be really easy to pick right way for specific destination. The other problem could be that routing software will not be able to properly guide you through the junction. If you take care that geometry of junction is represented correctly routing software will be able to guide you through the junction correctly. At least graphical representation should be correct. Question is, will (voice) instructions be correct? I guess that in such situation clever navigation software would avoid using instruction 'go straight', but would rather use instructions 'keep left' or 'keep right'. Blaz ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Clarifying and representing road markings at junctions
2009/9/30 Blaž Lorger blaz.lor...@triera.net: I don't see where problem lies. Is it that routing software will not be able to choose right route? You never stated it clearly, but if I understand correctly road from segment a to b has right of way over segment c. So all you need is a way to indicate this. There are some proposals that could solve this (stop signs, yield, right of way). If there are turn restrictions you can map those using turn restriction relation. But if there is no explicit right of way and no turn restrictions it really should not matter how road markings are painted on the road. Routing software should be able to pick the right route based on other criteria (road classification, speed limits, traffic calming, ...). obviously you will find those road markings in situations, where there are some kind of restrictions or explicit right of way. The thing is less to _find_ the best way, but to give apropriate indications (follow the street to the right or something similar, at least not simply: turn right. see this for explanations: http://www.atzl.eu/stickerei/images/stories/stickerei/f03.jpg http://www.phipsl.de/fall1.jpg http://cdn.fotocommunity.com/photos/10510230.jpgimgrefurl=http://www.fotocommunity.de/pc/pc/display/10510230usg=__A3IpXoI79f0MoS-wOtdSQLUJz3E=h=1000w=659sz=225hl=destart=12um=1tbnid=gU2vwxzupwOUYM:tbnh=149tbnw=98prev=/images%3Fq%3Dabknickende%2Bvorfahrt%26hl%3Dde%26client%3Dfirefox-a%26rls%3Dcom.ubuntu:de:unofficial%26sa%3DN%26um%3D1 cheers, Martin ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Clarifying and representing road markings at junctions
On Wednesday 30 September 2009 21:00:12 Martin Koppenhoefer wrote: 2009/9/30 Blaž Lorger blaz.lor...@triera.net: I don't see where problem lies. Is it that routing software will not be able to choose right route? You never stated it clearly, but if I understand correctly road from segment a to b has right of way over segment c. So all you need is a way to indicate this. There are some proposals that could solve this (stop signs, yield, right of way). If there are turn restrictions you can map those using turn restriction relation. But if there is no explicit right of way and no turn restrictions it really should not matter how road markings are painted on the road. Routing software should be able to pick the right route based on other criteria (road classification, speed limits, traffic calming, ...). obviously you will find those road markings in situations, where there are some kind of restrictions or explicit right of way. The thing is less to _find_ the best way, but to give apropriate indications (follow the street to the right or something similar, at least not simply: turn right. see this for explanations: http://www.atzl.eu/stickerei/images/stories/stickerei/f03.jpg http://www.phipsl.de/fall1.jpg http://cdn.fotocommunity.com/photos/10510230.jpgimgrefurl=http://www.fotoc ommunity.de/pc/pc/display/10510230usg=__A3IpXoI79f0MoS-wOtdSQLUJz3E=h=100 0w=659sz=225hl=destart=12um=1tbnid=gU2vwxzupwOUYM:tbnh=149tbnw=98p rev=/images%3Fq%3Dabknickende%2Bvorfahrt%26hl%3Dde%26client%3Dfirefox-a%26r ls%3Dcom.ubuntu:de:unofficial%26sa%3DN%26um%3D1 I was not able to open last one, but first two are cases where road with right of way is not the one going straight. By properly marking which way has right of way and making sure that junction geometry is correct, good navigation software should be able to produce sensible turn instructions without need for additional data. Blaz ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Clarifying and representing road markings at junctions
On Wed, Sep 30, 2009 at 7:35 PM, David Earl da...@frankieandshadow.com wrote: On 30/09/2009 10:20, Martin Koppenhoefer wrote: you could model it like this (see attached, colours are just indicating the ways, not highway-classes) Yes, that's also what I typically do, e.g. http://www.openstreetmap.org/?lat=52.596517lon=0.376144zoom=18layers=B000FTF Eek. Nice hack, but dodgy... 1) What if the road name changes *at* the junction, not just after the junction? 2) That hack just seems to change two things: a) it changes the *angle* between the intersecting ways at the junction. Is there any reason to want to do this? What exact problem does it solve? b) it makes a single *way* continue through the intersection. Does this actually infer that there is no giveway instruction? If so, is this documented anywhere? (I'm sure I could find examples where this is not the case) If not, then the hack *doesn't* explicitly show that the curved road continues through the intersection without interruption. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
[OSM-talk] Clarifying and representing road markings at junctions
I've been noticing recently a problem we're going to/already have in our data when it comes to routing directions particularly. It concerns how to define continuations of roads at junctions and/or the road markings that delineate that. This problem manifests itself in many ways but for a first example, look at the attached image (road.png). On the left you will see the physical plan of a road junction near where I live. The way that it would be represented in the OSM data model is shown on the right. In this case, it would be sensible to make a way out of the segments 'a' and 'b' (yeah, I know we don't have segments any more, it's just an explanation tool), call it, e.g. 'Curve Road' and make a second way out of segment 'c' and call it 'Small Road'. At this stage, the date representation is sound and routing application would have no problem knowing how to parse it. However, there are two (increasingly common) ways in which this model will be forced to be broken: 1. Naming doesn't match (e.g. [1]) This is the case near me. There used to be a road going along segment 'a' and 'c' called Frogmore Lane. Then when segment 'b' was built (and called Stonechat Road) they changed the road markings so that as you drive North from point 'A' they would guide you along 'b' towards 'B' (as in the left-hand picture). That is, you would be changed from being on Frogmore Lane to Stonechat Road without having 'turned'. Frogmore Lane continues along segment 'c'. In this case, I have to make 'a', 'b' and 'c' separate ways (well, 'a' and 'c' could be combined but that doesn't help) 2. Split for relations or some other property change Imagine a bus route goes along 'a' and 'b' while a cycle route goes along 'a' and 'c'. In order to place the correct ways in each relation, the three segments must be in separate ways. Topologically, this is just three ways meeting at a single node. There's no way to tell a driver to carry on along the road from A to B versus turn off the road at D along c. This information simply isn't in the database. Now, the routing application could try to guess the physical structure by looking at which two segments are most parallel but that would fail since the continuation is orthogonal to the road. They can't guess based on road names either due to point 1. Now in many ways I guess this is similar to the turn restrictions or street relations but they both have pitfalls when describing this sort of structure. I don't have a solution to this problem but I was hoping to spark a discussion about a simple and elegant way to describe this situation in those few (but frequent enough) places that this is necessary. Furthermore, Google directions frequently get this wrong! Thoughts/comments/suggestions? Regards, Matt Williams http://milliams.com [1] http://openstreetmap.org/?lat=50.911352lon=-1.016514zoom=18 attachment: road.png___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Clarifying and representing road markings at junctions
Hi, Matt Williams wrote: On the left you will see the physical plan of a road junction near where I live. The way that it would be represented in the OSM data model is shown on the right. [...] Your problem can and should be solved by a relation that models: to travel from node X to node Y on way(s) A,B,C the instruction to display is: . There are probably some clever algorithms to get some of these cases right but there will alway be cases that have to be modeled explicitly. The same kind of relation could also be used to describe signposting (to travel from X to Y on way(s) A, B, C follow the signs towards: ...). This is also knowledge that cannot be synthesized. Bye Frederik ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Clarifying and representing road markings at junctions
Hi, Roy Wallace wrote: In that case, use a relation. Two options: 1) http://wiki.openstreetmap.org/wiki/Relations/Proposed/Segmented_Tag [...] 2) http://wiki.openstreetmap.org/wiki/Relations/Proposed/Collected_Ways Both don't go far enough in my opinion. This is not a question of how can I express that two ways actually belong together, but the more general question of how can I model hints about the way in which the physical junction presents itself to the driver. For example you might have a junction that, in the OSM node/way representation, looks like a sideways T (i.e. it looks as if you can go straight on if you come from the South), and in reality it is Y shaped where if you come from the South you can either go half-left or half-right but never straight on. This, too, is a situation where you would want to model extra routing hints, and many others are possible. Bye Frederik ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Clarifying and representing road markings at junctions
On Wed, Sep 30, 2009 at 8:35 AM, Frederik Ramm frede...@remote.org wrote: Roy Wallace wrote: In that case, use a relation. Two options: 1) http://wiki.openstreetmap.org/wiki/Relations/Proposed/Segmented_Tag 2) http://wiki.openstreetmap.org/wiki/Relations/Proposed/Collected_Ways Both don't go far enough in my opinion. This is not a question of how can I express that two ways actually belong together, but the more general question of how can I model hints about the way in which the physical junction presents itself to the driver. I disagree. As Matt said, it is a question of how to define continuations of roads at junctions. Providing hints...to the driver, I think, is a job for routing software. For example you might have a junction that, in the OSM node/way representation, looks like a sideways T (i.e. it looks as if you can go straight on if you come from the South), and in reality it is Y shaped where if you come from the South you can either go half-left or half-right but never straight on. In this example, does the road continue without interruption on the left or right? If so, it can be modeled in the same way as Matt's original example. If not, shouldn't it just be modeled with 3 separate ways ending at the junction, as usual? Maybe I misunderstood the example - could you perhaps draw a picture? This, too, is a situation where you would want to model extra routing hints, and many others are possible. I don't think we should be storing routing hints in the OSM database. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk