
Hi list, hi Mark, First, a remark. Since the oneway=yes/cycleway=opposite roads have "(cycleway)" attached to their names, a GPS unit will randomly show either the regular name, or the (cycleway) name. Which isn't too bad for testing, I'd suggest you leave it this way until we're set and done with it, because now you can see *why* a certain road is accessible or inaccessible (i.e. it tells you where you're driving). Namely, here's one occasion where it came in handy: Yesterday I drove by car to one of those "cycleway=opposite" ways and to my surprise, my Garmin told me to turn right to "Hembrugstraat (cycleway)". I'm absolutely sure that the unit was set to "car" and not bike. So is there a bug in the opposite-way-code? (Or is this the strange idea of a Garmin that you can route the wrong way for some meters??) I had another Garmin unit with me with a regular Garmin map, that showed the right route; also, the route is nothing special, just about this: http://www.yournavigation.org/?flat=52.392997&flon=4.871082&tlat=52.391864&t... Any ideas? (I'll recheck the routing later on, to see if this will also happen with positions further away in one-way-streets). Best regards, Valentijn -- Durgerdamstraat 29, 1507 JL Zaandam; telefoon 075-7074579

Hi Valentijn,
First, a remark. Since the oneway=yes/cycleway=opposite roads have "(cycleway)" attached to their names, a GPS unit will randomly show either the regular name, or the (cycleway) name. Which isn't too bad for testing, I'd suggest you leave it this way until we're set and done with it, because now you can see *why* a certain road is accessible or inaccessible (i.e. it tells you where you're driving). Namely, here's one occasion where it came in handy:
Yes, that could be considered a "feature".
Yesterday I drove by car to one of those "cycleway=opposite" ways and to my surprise, my Garmin told me to turn right to "Hembrugstraat (cycleway)". I'm absolutely sure that the unit was set to "car" and not bike. So is there a bug in the opposite-way-code? (Or is this the strange idea of a Garmin that you can route the wrong way for some meters??) I had another Garmin unit with me with a regular Garmin map, that showed the right route; also, the route is nothing special, just about this: http://www.yournavigation.org/?flat=52.392997&flon=4.871082&tlat=52.391864&t...
Any ideas? (I'll recheck the routing later on, to see if this will also happen with positions further away in one-way-streets).
I think that (at least with mapsource) the routing restrictions are considered "advisory" at times rather than absolute prohibitions. If you start a route close to a road that has a synthesised cycleway I can quite believe that the gps would "grab" the cycleway instead of the road. If it routes down the cycleway from another way and a road is also available at the same point, that's not good. Cheers, Mark

Mark, Mark Burton schreef:
http://www.yournavigation.org/?flat=52.392997&flon=4.871082&tlat=52.391864&t... Any ideas? (I'll recheck the routing later on, to see if this will also happen with positions further away in one-way-streets). I think that (at least with mapsource) the routing restrictions are considered "advisory" at times rather than absolute prohibitions. If you start a route close to a road that has a synthesised cycleway I can quite believe that the gps would "grab" the cycleway instead of the road. If it routes down the cycleway from another way and a road is also available at the same point, that's not good.
In this case, it does: see the route above (the correct one); my Nuvi sends me right to the corner Spaarndammerdijk/Hembrugstraat and wants me to turn right to Hembrugstraat (cycleway). I know that accessibility is only advisory when there's no other way, but in this case, there's a clear route to the destination, but with the cycleway option turned on, it sends me in the wrong direction. I'll check what it does without the cycleways first. V.

Hi V, Yes, using mapsource, if the cycleway=opposite tag is present it will route a car into the last segment of Hembrugstraat using the cycleway but if the destination is not the last segment of that road or the cycleway tag is not present, it will route the car correctly. Experimentally, I split the last segment of that way to introduce a new new node and it behaved similarly i.e. it would route correctly right up to the last segment but when the dest was in the last segment it routed "arse about face" (as we say). Not at all sure where the problem is here - quite plausibly a bug in the Garmin routing engine. If so, perhaps we can work around it. Cheers, Mark

Hi Mark, Mark Burton schreef:
Yes, using mapsource, if the cycleway=opposite tag is present it will route a car into the last segment of Hembrugstraat using the cycleway but if the destination is not the last segment of that road or the cycleway tag is not present, it will route the car correctly.
OK. Hmm. What to check next. I suppose the cycleway, due to previous findings, is a copy of Hembrugstraat (i.e. not a real cycleway, but a regular road where cars are prohibited) with car traffic forbidden? Is there an easy way to make it a cycleway? Would that mean just removing the comments at //cycleWay.addTag("highway", "cycleway"); in the Osm5XmlHandler.java file? For now, I guess there's some sort of built-in rule that makes Garmin behave this way, but I'm not sure what it would be. Also, as a side note, wouldn't it be better to have: cycleWay.addTag("oneway", "yes"); instead of the current "no"? With "no", you'll never know on which road you're going to be, could be Anyroad or Anyroad (bicycle), while "yes" would give the Garmin no choice but put you on the Anyroad in the legal direction, while you'd be on Anyroad (bicycle) in the opposite direction? Or am I missing something? Best regards, Valentijn

I think the condition precedent you're thinking about is wrong here. The name that shows up in the routing does not have to be the street that is routed on. At least I see that behaviour quite frequently. How can I be sure? I use many non routable overlays and they have different naming. Nonetheless their names often show up in the route overview (gps and/or mapsource). This is using the "continue" patch, but I think the behaviour when I used my own adapted "make-opposite-cycleways" patch was the same.

Hi Mark, Without the all-cycleways option, it behaves normally. There must be some sort of Garmin rule in action here. I'll check if there's a better way (haha) to build the cycleway. Valentijn

On Fri, 2009-08-28 at 14:08 +0200, Valentijn Sessink wrote:
Hi Mark,
Without the all-cycleways option, it behaves normally. There must be some sort of Garmin rule in action here. I'll check if there's a better way (haha) to build the cycleway.
Routing involving cycleways (by bicycle or by car) makes me wonder why nobody's bothered getting Garmin's SDK and create new software for those units...

Paul Johnson schreef:
Routing involving cycleways (by bicycle or by car) makes me wonder why nobody's bothered getting Garmin's SDK and create new software for those units...
http://onroute.nl/ V. -- Durgerdamstraat 29, 1507 JL Zaandam; telefoon 075-7074579

Hi Valentijn, Paul, all, On Sat, Aug 29, 2009 at 07:29:08AM +0200, Valentijn Sessink wrote:
Paul Johnson schreef:
Routing involving cycleways (by bicycle or by car) makes me wonder why nobody's bothered getting Garmin's SDK and create new software for those units...
Can you please explain how that link is relevant to the Garmin SDK? It is a proprietary set of maps in Garmin format, right? (Do you happen to know which tools they are using in the map compilation?) Please correct and enlighten me if I am wrong, but I think that the only SDK that Garmin is distributing has something to do with a binary library for Microsoft Windows and Apple Mac OS X that speaks the proprietary Garmin protocols over USB or RS-232. Again, as far as I know, some recent models (Edge 605/705 at least) do not speak that protocol. Instead, they only implement the usb-storage driver, making the internal flash and the SD card accessible as block devices (drive letters in Windows parlance). I would love to hack the Garmin Edge 705 firmware to fix some annoyances, but I haven't been able to find anything about existing efforts. There is a firmware hacking scene for Canon digital cameras, for instance. Marko

Marko Mäkelä schreef:
On Sat, Aug 29, 2009 at 07:29:08AM +0200, Valentijn Sessink wrote:
http://onroute.nl/ Can you please explain how that link is relevant to the Garmin SDK?
It probably isn't. I know them for making a cycle map for the Netherlands. But they could be using anything to build their map. Valentijn

On Sat, 2009-08-29 at 07:29 +0200, Valentijn Sessink wrote:
Paul Johnson schreef:
Routing involving cycleways (by bicycle or by car) makes me wonder why nobody's bothered getting Garmin's SDK and create new software for those units...
What is it? (I don't speak Dutch and Google Translator wasn't much help).

Paul Johnson schreef:
On Sat, 2009-08-29 at 07:29 +0200, Valentijn Sessink wrote:
Paul Johnson schreef:
Routing involving cycleways (by bicycle or by car) makes me wonder why nobody's bothered getting Garmin's SDK and create new software for those units... http://onroute.nl/
What is it?
Sorry for the confusion. I know them for their bicycle map - they have a motorbike and walking map, too, it seems. Here's the Babelfish-translation of http://www.onroute.nl/fiets/112, edited by me to have one meaning. (I didn't fix the computer translation at all, I just made the meaning of the text clear, it's stil computer generated English). "By who were the OnRoute bicycle map developed? The OnRoute bicycle card were developed by a shop named WayPoint. WayPoint a shop is, specialised in leisure navigation. Started from the hobby, but nowadays we have three shops. Beside the main shop in Twentse Notter, there are two in Moordrecht (Gouda) and Vessem (Eindhoven). Waypoint sells GPS devices for a lot of years especially for walk and bicycles. But observed thereby that these devices for many people are too technical. This is what Waypoint has wanted to change with their OnRoute bicycle map. The map was made in association with the Belgian company RouteYou. RouteYou introduced especially technical mapping knowledge, which it made possible to connect our "beautiful roads network" to the maps of TeleAtlas. The conversion to Garmin-kaarten did WayPoint themselves. The "beautiful roads network" work came about on the basis of several sources, especially on the basis of the collected tracks of people who themselves cycled a certain route. Thus the site www.gpstracks.nl have for example made an important contribution. " Valentijn

It's nothing special. They are using standard garmin tools AFAICanTell. They learned from our maps that proper bicycle/topo maps will only provide good routing if preference is given in general for bicycleroutes. The autorouting is pretty good for cyclists, however because of 2 reasons: 1. their map data quality is very good for normal standards 2. Abundance of cycle routes exist in the netherlands. Making maps for the netherlands is easy, you just put heavy preference for cycleways and you get great routing. These maps are the first proprietary maps by Garmin that have good outdoor autorouting. (our efforts pay off in Garmin working harder/copying from us) Our main problem is as allways with OSM data unconnected streets (though much less than some month ago as after months of complaints potlatch got better in this regard), and too sharp turns (this would need to be solved by preprocessing of osm data befote compiling with mkgmap) . Garmin seems to straighten curvy sections (I know that they do this for the topo Swiss, which frequently on very curvy small mountain trails shows too short distance, in my eyes done on purpose so that such routes are chosen even though garmin gps penalise any curve/corner. This straightened data is only used in the NOD sections. the ways themselves as they show on the map are longer than the underlying routing data. I don't know whether onroute maps have bin built using the same principle, but this could be well possible. Felix 2009/8/29 Valentijn Sessink <valentyn@blub.net>
Paul Johnson schreef:
On Sat, 2009-08-29 at 07:29 +0200, Valentijn Sessink wrote:
Paul Johnson schreef:
Routing involving cycleways (by bicycle or by car) makes me wonder why nobody's bothered getting Garmin's SDK and create new software for those units... http://onroute.nl/
What is it?
Sorry for the confusion. I know them for their bicycle map - they have a motorbike and walking map, too, it seems.
Here's the Babelfish-translation of http://www.onroute.nl/fiets/112, edited by me to have one meaning. (I didn't fix the computer translation at all, I just made the meaning of the text clear, it's stil computer generated English).
"By who were the OnRoute bicycle map developed?
The OnRoute bicycle card were developed by a shop named WayPoint. WayPoint a shop is, specialised in leisure navigation. Started from the hobby, but nowadays we have three shops. Beside the main shop in Twentse Notter, there are two in Moordrecht (Gouda) and Vessem (Eindhoven).
Waypoint sells GPS devices for a lot of years especially for walk and bicycles. But observed thereby that these devices for many people are too technical. This is what Waypoint has wanted to change with their OnRoute bicycle map.
The map was made in association with the Belgian company RouteYou. RouteYou introduced especially technical mapping knowledge, which it made possible to connect our "beautiful roads network" to the maps of TeleAtlas. The conversion to Garmin-kaarten did WayPoint themselves. The "beautiful roads network" work came about on the basis of several sources, especially on the basis of the collected tracks of people who themselves cycled a certain route. Thus the site www.gpstracks.nl have for example made an important contribution. "
Valentijn _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

Interesting information. Can I ask how you know that Garmin learned from OSM how to make cyclemaps? Some things I learned about the Garmin way of doing business: I was on the sideline watching when the Dutch OSM chapter was talking with Waypoint about joining forces and make e.g. cyclemaps using OSM data (they sell more devices and free maps, we get a bigger userbase). But I guess Garmin pulled a string because OSM was suddenly not a possibility anymore. I also learned from the owner of my default bikeshop that when he started selling Garmin devices, he was pushed heavily to also sell certain amounts of Garmin maps. The shop owner wanted to be able to advice independently so he stopped selling the devices. Felix Hartmann wrote:
It's nothing special. They are using standard garmin tools AFAICanTell. They learned from our maps that proper bicycle/topo maps will only provide good routing if preference is given in general for bicycleroutes.
The autorouting is pretty good for cyclists, however because of 2 reasons: 1. their map data quality is very good for normal standards 2. Abundance of cycle routes exist in the netherlands.
Making maps for the netherlands is easy, you just put heavy preference for cycleways and you get great routing. These maps are the first proprietary maps by Garmin that have good outdoor autorouting. (our efforts pay off in Garmin working harder/copying from us)
Our main problem is as allways with OSM data unconnected streets (though much less than some month ago as after months of complaints potlatch got better in this regard), and too sharp turns (this would need to be solved by preprocessing of osm data befote compiling with mkgmap) . Garmin seems to straighten curvy sections (I know that they do this for the topo Swiss, which frequently on very curvy small mountain trails shows too short distance, in my eyes done on purpose so that such routes are chosen even though garmin gps penalise any curve/corner. This straightened data is only used in the NOD sections. the ways themselves as they show on the map are longer than the underlying routing data. I don't know whether onroute maps have bin built using the same principle, but this could be well possible.
Felix
2009/8/29 Valentijn Sessink <valentyn@blub.net <mailto:valentyn@blub.net>>
Paul Johnson schreef: > On Sat, 2009-08-29 at 07:29 +0200, Valentijn Sessink wrote: >> Paul Johnson schreef: >>> Routing involving cycleways (by bicycle or by car) makes me wonder why >>> nobody's bothered getting Garmin's SDK and create new software for those >>> units... >> http://onroute.nl/ > > What is it?
Sorry for the confusion. I know them for their bicycle map - they have a motorbike and walking map, too, it seems.
Here's the Babelfish-translation of http://www.onroute.nl/fiets/112, edited by me to have one meaning. (I didn't fix the computer translation at all, I just made the meaning of the text clear, it's stil computer generated English).
"By who were the OnRoute bicycle map developed?
The OnRoute bicycle card were developed by a shop named WayPoint. WayPoint a shop is, specialised in leisure navigation. Started from the hobby, but nowadays we have three shops. Beside the main shop in Twentse Notter, there are two in Moordrecht (Gouda) and Vessem (Eindhoven).
Waypoint sells GPS devices for a lot of years especially for walk and bicycles. But observed thereby that these devices for many people are too technical. This is what Waypoint has wanted to change with their OnRoute bicycle map.
The map was made in association with the Belgian company RouteYou. RouteYou introduced especially technical mapping knowledge, which it made possible to connect our "beautiful roads network" to the maps of TeleAtlas. The conversion to Garmin-kaarten did WayPoint themselves. The "beautiful roads network" work came about on the basis of several sources, especially on the basis of the collected tracks of people who themselves cycled a certain route. Thus the site www.gpstracks.nl <http://www.gpstracks.nl> have for example made an important contribution. "
Valentijn _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk <mailto:mkgmap-dev@lists.mkgmap.org.uk> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
------------------------------------------------------------------------
_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

Lambertus wrote:
Interesting information. Can I ask how you know that Garmin learned from OSM how to make cyclemaps?
Well the latest (I think it came out 4-5 weeks ago) update for the onroute cycle map suddenly provided much better routing than anything Garmin / Garmin subsidiaries had produced before. Analysing the road_class ( road_speed they set equal more or less for all roads) they used the same principles as Thilo and me when it comes to prefer cycleroutes. Also they did leave out routing information for motorways (noticing as we did that the restrictions don't work 100%). So in effect just like Thilo's and my maps, the onroute fietskart has best autorouting when used with car/motorcycle setting. This seems to be a bit of a coincidence why this happens now once that once good outdoor autorouting appears they for the first time in over 10 years of Garmin topographic maps come out with topo maps that actually do have usable autorouting. Up to now it was good enough to use it over a distance of say max 5 junctions to the next waypoint, so having the only advantage of being quicker to create autourouted routes over tracks. Of course this approach will only work in the netherlands, and in other countries they will have big problems getting to the same standards, as 1. there exist fewer cycleways and 2. they don't have good enough attributes available so in the end we at OSM will continue to be able to provide maps that get the same standards without using OSM data (which luckily they are not allowed to mix with proprietary map data). Garmin must be well aware off the OSM scene by now. On openmtbmap.org I have daily around 500 people downloading maps (Germany alone 100-200 downloads a day). Computerteddy's maps probabely lead the pack being the first ones to be available for any region in Europe. Others also get many downloads. On mtb-news.de (the biggest mtb forum worldwide) in the GPS section the topic about openmtbmaps has after not yet 5month around 58.000 hits and 862 comments (compared to 18.000 hits on the topic about the Garmin Topo Deutschland v3 which is 2 much older). I don't have a clue how often the Topo Deutschland v3 is actually sold, but I bet that if you exclude users who bought it with their device, and exclude users who used a keygen/garmin unlocker to access this map, there are more users of OSM Germany maps already. Of course important decisions at Garmin are taken in the States where OSM community is still unimportant, but I am sure that in the course of the next year Garmin will react somehow on OSM.
Some things I learned about the Garmin way of doing business: I was on the sideline watching when the Dutch OSM chapter was talking with Waypoint about joining forces and make e.g. cyclemaps using OSM data (they sell more devices and free maps, we get a bigger userbase). But I guess Garmin pulled a string because OSM was suddenly not a possibility anymore.
Yeah I know that many people at Garmin flirted with the idea of dropping map creation alltogether because of too many people using keygens to crack their maps and too high effort to produce maps. This stopped abruptly around 2 month ago.
I also learned from the owner of my default bikeshop that when he started selling Garmin devices, he was pushed heavily to also sell certain amounts of Garmin maps. The shop owner wanted to be able to advice independently so he stopped selling the devices.
Felix Hartmann wrote:
It's nothing special. They are using standard garmin tools AFAICanTell. They learned from our maps that proper bicycle/topo maps will only provide good routing if preference is given in general for bicycleroutes.
The autorouting is pretty good for cyclists, however because of 2 reasons: 1. their map data quality is very good for normal standards 2. Abundance of cycle routes exist in the netherlands.
Making maps for the netherlands is easy, you just put heavy preference for cycleways and you get great routing. These maps are the first proprietary maps by Garmin that have good outdoor autorouting. (our efforts pay off in Garmin working harder/copying from us)
Our main problem is as allways with OSM data unconnected streets (though much less than some month ago as after months of complaints potlatch got better in this regard), and too sharp turns (this would need to be solved by preprocessing of osm data befote compiling with mkgmap) . Garmin seems to straighten curvy sections (I know that they do this for the topo Swiss, which frequently on very curvy small mountain trails shows too short distance, in my eyes done on purpose so that such routes are chosen even though garmin gps penalise any curve/corner. This straightened data is only used in the NOD sections. the ways themselves as they show on the map are longer than the underlying routing data. I don't know whether onroute maps have bin built using the same principle, but this could be well possible.
Felix
2009/8/29 Valentijn Sessink <valentyn@blub.net <mailto:valentyn@blub.net>>
Paul Johnson schreef: > On Sat, 2009-08-29 at 07:29 +0200, Valentijn Sessink wrote: >> Paul Johnson schreef: >>> Routing involving cycleways (by bicycle or by car) makes me wonder why >>> nobody's bothered getting Garmin's SDK and create new software for those >>> units... >> http://onroute.nl/ > > What is it?
Sorry for the confusion. I know them for their bicycle map - they have a motorbike and walking map, too, it seems.
Here's the Babelfish-translation of http://www.onroute.nl/fiets/112, edited by me to have one meaning. (I didn't fix the computer translation at all, I just made the meaning of the text clear, it's stil computer generated English).
"By who were the OnRoute bicycle map developed?
The OnRoute bicycle card were developed by a shop named WayPoint. WayPoint a shop is, specialised in leisure navigation. Started from the hobby, but nowadays we have three shops. Beside the main shop in Twentse Notter, there are two in Moordrecht (Gouda) and Vessem (Eindhoven).
Waypoint sells GPS devices for a lot of years especially for walk and bicycles. But observed thereby that these devices for many people are too technical. This is what Waypoint has wanted to change with their OnRoute bicycle map.
The map was made in association with the Belgian company RouteYou. RouteYou introduced especially technical mapping knowledge, which it made possible to connect our "beautiful roads network" to the maps of TeleAtlas. The conversion to Garmin-kaarten did WayPoint themselves. The "beautiful roads network" work came about on the basis of several sources, especially on the basis of the collected tracks of people who themselves cycled a certain route. Thus the site www.gpstracks.nl <http://www.gpstracks.nl> have for example made an important contribution. "
Valentijn _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk <mailto:mkgmap-dev@lists.mkgmap.org.uk> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
------------------------------------------------------------------------
_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

Mark, I've tried a couple of different methods to build opposite-cycleways, namely forbid all motoring traffic, make a real cycleway, but all of them make the Garmin turn to the last part of the oneway Hembrugstraat. So I conclude it's a Garmin issue. I'm sure we could come up with all sorts of nasty tricks to work around it, but I'd suggest we leave it this way. Valentijn

Valentijn,
I've tried a couple of different methods to build opposite-cycleways, namely forbid all motoring traffic, make a real cycleway, but all of them make the Garmin turn to the last part of the oneway Hembrugstraat. So I conclude it's a Garmin issue. I'm sure we could come up with all sorts of nasty tricks to work around it, but I'd suggest we leave it this way.
OK, thanks for investigating that. For my own part, I tried pruning the last segment from the synthesised cycleway and, as expected, it stopped the bad behaviour in that the car routing never tried to do the right turn. However, as expected, it broke the cycle routing through the last segment. So I agree, we should leave it as is for the moment. Cheers, Mark

Mark, I think the following could be what happens. The GPS unit tries to route you to $dest, which is on a non accessible road. As "non accessible" has no access, it makes no sense trying other directions. If there's a crossing, that's a reason to try other directions so the oneway street will come up. I was thinking: are the oneway street and the fake route interconnected automatically? I know they share the nodes, is that the same as being able to skip between them? I.e. is it: | | +--+--------+-----+----+---- a street +--+--------+-----+----+---- a street (cycleway) | | ... or | | +--.--------.-----+----.---- a street +--.--------.-----+----.---- a street (cycleway) | | ? Valentijn -- Durgerdamstraat 29, 1507 JL Zaandam; telefoon 075-7074579

Valentijn,
I was thinking: are the oneway street and the fake route interconnected automatically? I know they share the nodes, is that the same as being able to skip between them?
I.e. is it:
| | +--+--------+-----+----+---- a street +--+--------+-----+----+---- a street (cycleway) | |
... or
| | +--.--------.-----+----.---- a street +--.--------.-----+----.---- a street (cycleway) | |
Effectively, it's the later. Although the cycleway and the road share the same points, unless a point is already a routing node because some other way shares the point, it will not be made into a routing node. You don't want the former because you would create many unwanted routeing nodes. Cheers, Mark
participants (6)
-
Felix Hartmann
-
Lambertus
-
Mark Burton
-
Marko Mäkelä
-
Paul Johnson
-
Valentijn Sessink