Re: [talk-au] traffic lights on dual carriageway intersections
So, Ian Sergeant has presented reasoning why we should not pursue more complicated schemes for applying traffic lights to intersections of dual carriageways - fair enough. This brings me back to the incident that triggered me to start this thread: there are several intersections of dual carriageways in Perth CBD where only 1 of the 4 nodes are marked with traffic lights, and this struck me as wrong, and hence I asked what was the correct and accepted method. If we are to reject the more complex solution of adding traffic lights one node back from the interesting nodes (as implemented in Melbourne CBD, and reasoned against by Ian), surely we should be marking all 4 intersection nodes as having traffic lights ?? (not just one). what does everyone think ? Ian Steer ___ Talk-au mailing list Talk-au@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] traffic lights on dual carriageway intersections
On 07/11/12 23:21, Steer wrote: So, Ian Sergeant has presented reasoning why we should not pursue more complicated schemes for applying traffic lights to intersections of dual carriageways -- fair enough. That is not quite what I said. I'd be happy to see a more detailed schema that is expressive enough to indicate where the stop line is, the physical location of the signals, which signals are in sync, how many signals on a journey, etc. The current one tag/independent node system means that you need to make a choice in what you can represent. Since I can't see a way to generally and accurately represent traffic light count in the current schema, I think that is the wrong choice to represent just on dual carriageways. I think a relation that links these nodes is probably inevitable. Ian. ___ Talk-au mailing list Talk-au@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] traffic lights on dual carriageway intersections
By choosing to place traffic light not on the intersection node, you are failing to represent that this is an intersection of two roads, controlled by traffic signals. I don't see how it is failing to represent that - the intersection is there (the ways intersect at nodes), and there are traffic signals *before* the intersection (not smack-bang in the middle of the intersection) Instead you are choosing to represent There is a stop line here and traffic signal and further on there is an intersection. - but isn't that EXACTLY what we have - a stop line with a traffic signal, with an intersection further on ? - and if we were REALLY keen, the same *could* be done for single carriage way intersections (but I'm not suggesting that that is a sensible option) Ian ___ Talk-au mailing list Talk-au@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] traffic lights on dual carriageway intersections
On 4 November 2012 20:58, Ian Steer ianst...@iinet.net.au wrote: By choosing to place traffic light not on the intersection node, you are failing to represent that this is an intersection of two roads, controlled by traffic signals. I don't see how it is failing to represent that - the intersection is there (the ways intersect at nodes), and there are traffic signals *before* the intersection (not smack-bang in the middle of the intersection) Because our current schema indicates that an intesection is controlled by signals by placing the traffic signals on the intersecting node. Traffic lights do occur before intersections or the immediate vicinity without controlling traffic movements through that intersection. It is a meaningful respresentation. In many North American cities the signals hang right in the centre of the intersection. Are you saying these should be mapped differently just because the lights are located in a different location? As far as the road user is concerned, they are the same. They don't care where the traffic signals are - they care there are lights at the corner of 6th and Vine. - and if we were REALLY keen, the same *could* be done for single carriage way intersections (but I'm not suggesting that that is a sensible option) Exactly. This is the clincher. Why on earth would you develop a schema that is only relevant to dual carriageways? When there is a schema that can respresent stop lines, signal locations, and intersection control across all junctions then I'm in. Until then, trying to vary the current schema in a way that is both ambigious, and only works for dual carriageways just doesn't fly, IMO. Ian. ___ Talk-au mailing list Talk-au@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] traffic lights on dual carriageway intersections
Ian Steer wrote I think this is good because no matter which way you go through the intersection, you only pass one set of lights (rather than 2 if they were placed on the actual intersecting nodes). Couldn't a smart traffic light counter detect dual carrageways and just add a single signal, same as the exit counter does for roundabouts? Nick ___ Talk-au mailing list Talk-au@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] traffic lights on dual carriageway intersections
Steer wrote: I have been trying to find the accepted practise for mapping traffic lights where dual carriageways interest. There is much discussion on various sites, but most seems to be a bit old, and I’m not convinced I’ve found what is the latest accepted practise. I checked some intersections in Melbourne’s CBD, and the method I saw that I liked and thought the best was where there were 4 lights at the intersection, but they were not placed on the intersecting modes, but one node back “upstream” on each way. I think this is good because no matter which way you go through the intersection, you only pass one set of lights (rather than 2 if they were placed on the actual intersecting nodes). Any comments? I have always entered such traffic lights on dual carriageways in the way you describe. This is because: 1. The traffic light count along a section of road is then accurate, and 2. It's the accurate representation of what's on the ground. It lets us convey the significance of the stop lines associated with the lights. That's something we can't do with two-way traffic without compromising point 1. I have argued this position on previous occasions. John ___ Talk-au mailing list Talk-au@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] traffic lights on dual carriageway intersections
And the only area it's done like this is in Melbourne. Cheers Ross On 03/11/12 17:03, John Henderson wrote: Steer wrote: I have been trying to find the accepted practise for mapping traffic lights where dual carriageways interest. There is much discussion on various sites, but most seems to be a bit old, and I’m not convinced I’ve found what is the latest accepted practise. I checked some intersections in Melbourne’s CBD, and the method I saw that I liked and thought the best was where there were 4 lights at the intersection, but they were not placed on the intersecting modes, but one node back “upstream” on each way. I think this is good because no matter which way you go through the intersection, you only pass one set of lights (rather than 2 if they were placed on the actual intersecting nodes). Any comments? I have always entered such traffic lights on dual carriageways in the way you describe. This is because: 1. The traffic light count along a section of road is then accurate, and 2. It's the accurate representation of what's on the ground. It lets us convey the significance of the stop lines associated with the lights. That's something we can't do with two-way traffic without compromising point 1. I have argued this position on previous occasions. John ___ Talk-au mailing list Talk-au@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-au ___ Talk-au mailing list Talk-au@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] traffic lights on dual carriageway intersections
On 3/11/2012 5:33 PM, John Henderson wrote: I checked some intersections in Melbourne’s CBD, and the method I saw that I liked and thought the best was where there were 4 lights at the intersection, but they were not placed on the intersecting modes, but one node back “upstream” on each way. I think this is good because no matter which way you go through the intersection, you only pass one set of lights (rather than 2 if they were placed on the actual intersecting nodes). I read it and liked it but then poked around near me but found that traffic signals where a divided road meets and undivided road, the undivided road gets a count of two. You could put the undivided (two ray) road traffic signals in the centre of the intersection but that starts to look pretty strange. Which then leads us to possible accusations of mapping for the routing renderer. Strictly speaking the traffic lights are things on poles placed on traffic islands as well as overhead gantries. Should we be tagging the physical object, ie. the signal rather than its effect which is most pronounced at the stop-line? Another thought would be to tag the stopline with a direction tag to hint the renderer that a vehicle would stop here moving in a particular direction..starts to get complicated. What about wig-wags outside fire-stations or supplementary traffic signals applied to a level crossing. Starts to get tricky.. Still worth thinking about... Alex ___ Talk-au mailing list Talk-au@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] traffic lights on dual carriageway intersections
On 03/11/12 18:03, John Henderson wrote: 2. It's the accurate representation of what's on the ground. It lets us convey the significance of the stop lines associated with the lights. That's something we can't do with two-way traffic without compromising point 1. Mapping is choosing a representation of what is on the ground. By choosing to place traffic light not on the intersection node, you are failing to represent that this is an intersection of two roads, controlled by traffic signals. Instead you are choosing to represent There is a stop line here and traffic signal and further on there is an intersection. So, ideally we should have a rich enough mapping set to allow us to indicate both. However, since we can currently represent only one, I currently feel that it is far more important to indicate that the intersection is controlled, than the location of the traffic signals, or an accurate count of traffic signals. This is especially true, since in the general case (non-dual carriageway) we can't represent these things anyway. So, even if we favour the stop line location/traffic signal count method, it will always be wrong and unreliable. Ian. ___ Talk-au mailing list Talk-au@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-au
Re: [talk-au] traffic lights on dual carriageway intersections
On 04/11/12 07:29, Ian Sergeant wrote: By choosing to place traffic light not on the intersection node, you are failing to represent that this is an intersection of two roads, controlled by traffic signals. Instead you are choosing to represent There is a stop line here and traffic signal and further on there is an intersection. We have different intuitions about what's important here. John ___ Talk-au mailing list Talk-au@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-au
[talk-au] traffic lights on dual carriageway intersections
I have been trying to find the accepted practise for mapping traffic lights where dual carriageways interest. There is much discussion on various sites, but most seems to be a bit old, and I'm not convinced I've found what is the latest accepted practise. I checked some intersections in Melbourne's CBD, and the method I saw that I liked and thought the best was where there were 4 lights at the intersection, but they were not placed on the intersecting modes, but one node back upstream on each way. I think this is good because no matter which way you go through the intersection, you only pass one set of lights (rather than 2 if they were placed on the actual intersecting nodes). Any comments? thanks Ian ___ Talk-au mailing list Talk-au@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-au