Re: [Talk-ca] Routing tool for openstreetmap.ca?
So is there a way to 'teach' that better routes? Blairmore to Calgary was routed through Fort McLeod (257km)... when the faster/shorted route is via Highway 22 and 533 across to Nanton (217km). It might say something about my driving, but that would take a little over 2hrs rather than suggested 3hr4. Yes, I average more that 70km/hr... I'm not familiar with either route, or with your driving style. :-) As you suggested there may be an issue with the 'default speed' on these ways, but I don't think it is sensible to tag every segment with a speed limit when it is 'generic'. Is there a way to inspect how the router 'sees' the roads? The route follows paved rural primary roads (100km/hr and 110km/hr speed limits) and thus the average speed of 70km/hr seems way off traffic flow is more like 120km/hr. Simon ___ Talk-ca mailing list Talk-ca@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-ca
Re: [Talk-ca] Routing tool for openstreetmap.ca?
On Mon, Apr 22, 2013 at 12:30 PM, Samuel Longiaru longi...@shaw.ca wrote: Another big thanks here to those involved in setting this up! I do have a suggestion for the site. Perhaps it is already implemented elsewhere, in which case maybe all I need is to be reminded of its location so I can update my bookmarks. I think it would be great to have access to a routing engine using as current a Canadian data extract as possible... like daily or even more recent. http://map.project-osrm.org/ For Canadian data and the rest of the world. Updates the data twice a day, as I understand it. ___ Talk-ca mailing list Talk-ca@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-ca
Re: [Talk-ca] Routing tool for openstreetmap.ca?
http://map.project-osrm.org/ For Canadian data and the rest of the world. Updates the data twice a day, as I understand it. So is there a way to 'teach' that better routes? Blairmore to Calgary was routed through Fort McLeod (257km)... when the faster/shorted route is via Highway 22 and 533 across to Nanton (217km). It might say something about my driving, but that would take a little over 2hrs rather than suggested 3hr4. Yes, I average more that 70km/hr... Simon ___ Talk-ca mailing list Talk-ca@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-ca
Re: [Talk-ca] Routing tool for openstreetmap.ca?
On Mon, Apr 22, 2013 at 1:51 PM, si...@mungewell.org wrote: http://map.project-osrm.org/ For Canadian data and the rest of the world. Updates the data twice a day, as I understand it. So is there a way to 'teach' that better routes? Blairmore to Calgary was routed through Fort McLeod (257km)... when the faster/shorted route is via Highway 22 and 533 across to Nanton (217km). It might say something about my driving, but that would take a little over 2hrs rather than suggested 3hr4. Yes, I average more that 70km/hr... I'm not familiar with either route, or with your driving style. :-) Routers using OSM data will make assumptions where speed limit data in not available so you might be running into issues where the assumptions don't match your driving experience on the ground. In past, I've found that there are connectivity problems in the OSM data, when routers make suggestions taht I wouldn't expect. In fact, that was one of the things we were using the test instance of OSRM for; finding discontinuities, bad one-ways, and other tagging / mapping errors. ___ Talk-ca mailing list Talk-ca@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-ca
Re: [Talk-ca] Routing tool for openstreetmap.ca?
- Original Message - From: Richard Weait rich...@weait.com To: Simon Wood si...@mungewell.org Cc: Talk-CA OpenStreetMap talk-ca@openstreetmap.org Sent: Mon, 22 Apr 2013 12:14:42 -0600 (MDT) Subject: Re: [Talk-ca] Routing tool for openstreetmap.ca? On Mon, Apr 22, 2013 at 1:51 PM, si...@mungewell.org wrote: http://map.project-osrm.org/ For Canadian data and the rest of the world. Updates the data twice a day, as I understand it. So is there a way to 'teach' that better routes? Blairmore to Calgary was routed through Fort McLeod (257km)... when the faster/shorted route is via Highway 22 and 533 across to Nanton (217km). It might say something about my driving, but that would take a little over 2hrs rather than suggested 3hr4. Yes, I average more that 70km/hr... I'm not familiar with either route, or with your driving style. :-) Routers using OSM data will make assumptions where speed limit data in not available so you might be running into issues where the assumptions don't match your driving experience on the ground. In past, I've found that there are connectivity problems in the OSM data, when routers make suggestions taht I wouldn't expect. In fact, that was one of the things we were using the test instance of OSRM for; finding discontinuities, bad one-ways, and other tagging / mapping errors. Richard, Thanks for the link to the OSRM site, but I don't think that was it. I'm familiar with that project and recently have been following the dev list. At least as the demo site stands, it does give crazy routings for the area of South Australia where I am right now. And while it may be related to speed limit tags, it's not for the lack of them, but because they exist. The unpaved roads here have been correctly tagged with max_speed=100. While that is the statutory limit for unsigned rural roads in South Australia, it is not reasonable in practice. When OSRM sees that, it coughs up routes that suggest one exit reasonably good motorways and jump onto unpaved roads. Here's a good example: http://osrm.at/2Vl When routing from Adelaide to the Roseworthy Campus, OSRM routes one off the M20 and onto unpaved roads when staying on the M20 for one more exit, then exiting to Redbanks Road is the much more logical choice. The fact that OSRM updates data twice a day is encouraging. I didn't see that anywhere. But I've found that the Gosmore engine, at least as implemented by http://yournavigation.org makes more reasonable assumptions and so comes up with more logical routes. The problem there, however is that yournavigation appears to be using worldwide data over 2 months old. And hence my suggestion for an implementation of a routing feature that uses reasonable assumptions (or as best as we can agree to) and utilizing the latest Canadian data possible. It could be based on either the OSRM engine or the Gosmore engine I would imagine. It seems to me that the OSRM routing could benefit greatly by a 0.6 penalty for unpaved roads as had been suggested a few time before, but they don't seem to want to go that way. Thanks, Sam ___ Talk-ca mailing list Talk-ca@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-ca
Re: [Talk-ca] Routing tool for openstreetmap.ca?
On Mon, Apr 22, 2013 at 3:33 PM, Samuel Longiaru longi...@shaw.ca wrote: It seems to me that the OSRM routing could benefit greatly by a 0.6 penalty for unpaved roads as had been suggested a few time before, but they don't seem to want to go that way. Why incur a penalty just because the roadway is unpaved? A better solution would be to have the ability to request paved roads only when routing. That way the user could decide whether an unpaved roadway should be selected or not. I suppose the best solution would be to allow the user to select whether unpaved roads can be used for routing, and also allow the user to select the penalty to apply for unpaved. I fight with my GPS all the time. I tell it to never use unpaved roads, but it will put me onto them quite often. Then on the other hand it can try and send me on long detours some times when I know I want to take that 2 mile shortcut on gravel to save 40 miles on pavement. It's pretty tough to teach a computer to be as wishy-washy as a human! -- James VE6SRV ___ Talk-ca mailing list Talk-ca@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-ca
Re: [Talk-ca] Routing tool for openstreetmap.ca?
- James Ewen ve6...@gmail.com wrote: On Mon, Apr 22, 2013 at 3:33 PM, Samuel Longiaru longi...@shaw.ca wrote: It seems to me that the OSRM routing could benefit greatly by a 0.6 penalty for unpaved roads as had been suggested a few time before, but they don't seem to want to go that way. Why incur a penalty just because the roadway is unpaved? A better solution would be to have the ability to request paved roads only when routing. That way the user could decide whether an unpaved roadway should be selected or not. I suppose the best solution would be to allow the user to select whether unpaved roads can be used for routing, and also allow the user to select the penalty to apply for unpaved. I fight with my GPS all the time. I tell it to never use unpaved roads, but it will put me onto them quite often. Then on the other hand it can try and send me on long detours some times when I know I want to take that 2 mile shortcut on gravel to save 40 miles on pavement. It's pretty tough to teach a computer to be as wishy-washy as a human! -- James VE6SRV ___ Talk-ca mailing list Talk-ca@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-ca Paved roads only option? Yes, that's one way. Essentially giving unpaved roads a penalty value (factor) of 0. Then unpaved roads wouldn't be routed on. But consider the case where you are on an unpaved road and wish to route somewhere using the 'avoid unpaved roads' option. It would seem to me that in that case, the engine will need to assess a reasonable penalty for unpaved roads and minimize that penalty by getting you to paved roads by the quickest or shortest means. So either way you stack it, at some point, you need to assign a penalty. Right now, on the OSRM site, you can neither assign a penalty nor elect a paved-roads only option. All roads are reated equally. The yournavigation site must be doing something different, as it yields different (and more logical) results. I'd love to see a routing engine with a desireability factor that could be adjustable. If you really loathe unpaved roads, you could set the unpaved roads desirability factor low (i.e., apply a greater penalty for unpaved roads). And if you don't really care all that much whether you take paved or unpaved, then set the factor high. If your GPS had that, then maybe you wouldn't be fighting with it so much. :) Sam ___ Talk-ca mailing list Talk-ca@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-ca
Re: [Talk-ca] Routing tool for openstreetmap.ca?
On Mon, Apr 22, 2013 at 7:57 PM, Samuel Longiaru longi...@shaw.ca wrote: If your GPS had that, then maybe you wouldn't be fighting with it so much. :) Or if the database contained road surface information, proper speed limit data, and other valuable information, then the routing engine would have a chance at knowing where to send you. It's a challenge to determine whether the routing engine or the database is to blame for the routing choices. With OSM, we have access to the database, and only ourselves to blame if the information required is not in the database! :) -- James VE6SRV ___ Talk-ca mailing list Talk-ca@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-ca
