Road Types (USA) - Walking Trails

[AlanOfTheBerg]

I would leave them alone unless they are actually causing problems. My guess is they're acting like railroads act now, and are essentially invisible to the routing server. My opinion is that they look quite nice

I verified today that Waze will still route vehicles over connected walking paths. So if they are connected they are definitely not invisible; in my test, Waze routed me from a street over a connected walking path and thence onto a connected parking-lot road.

I verified this a while ago, too. Walking Trails, specifically. I did not test other non-drivables though. Waze treats Walking Trail like a PLot or Private Dr in that it will avoid them as through routes, but will route onto them for destination or use them as a starting point. That is starkly different from Railroads, which the routing server appears to ignore completely.

[CBenson]

I have not been able to figure out a consistent behavior for walking trails. Addresses I've tested sometimes route to the end of the trail, then don't, then do again. Waze has never to my knowledge really explained what is the intended behavior for walking trails.

This is very difficult for me to test, as address searches in general have recently changed for me. They now route to the street in the address for me in the client.

My standby test has been to 1600 Ritchie Hwy, Arnold, MD 21012 as discussed here. Currently due to waze actually using the street name as the destination, the client gives me a route to Ritchie Hwy even though both Baltimore Annapolis Blvd and the B & A Trail are closer to the pin. However, the live map routes some where near the end of the trail.

I don't see as many issues reported with walking trails anymore. But some failures to get any route still seem to occur when the route is near a long walking trail. If I try to route to 1820 Wiehle Ave, Reston, VA, I can't get a route in the live map as discussed here. But the client now routes me to a point on Wiehle Ave.

I think I remember that at one time waze stated that intended the routing to work like the diagram in the wiki. But I'm not sure I've ever actually seen it work that way. Rather, the routing (when the walking trail effects the routing) seems to be to some point near the end of the walking trail, regardless of whether other segments come close (or ever cross) the walking trail in the middle.

This walking trail behavior has been unpredictable. I know others have attempted to use it as you have to get good routing to parking lots without success. I wish we could predict the behavior of walking trails better and use them to some benefit.

[CBenson]

How they behave if completely unconnected, I haven't checked. If they were as bad as Alan's tests from a few months ago suggested, I would have thought we'd have URs everywhere by now. So I'm lost as to just how dangerous the disconnected ones are.

I have dealt with URs within the past week or so where unconnected walking trails caused routing issues. For the trail I typically test, I get the same results I have for a while. This route is effected by the walking trail. If you move the destination pin up near Ashcroft Ct, but still nearer to the trail, then waze can't return a route at all. What has changed for me is how waze routes to Google address results in the client, making this issue much more difficult to reproduce in the client. There are much fewer circumstances where waze is simply routing to the closest segment it can find.

[CBenson]

But disconnected walking trails can effect routing. So this may not be enough. We need to understand how walking trails and pedestrian boardwalks effect searching and routing.

[dougp01]

I have had a similar problem and my search of the wiki brought me to this thread. I am now experimenting with a walking trail of my own. Interesting to note: Walking Trails are in the "Non-drivable" category, so I obviously assumed it would be safe to integrate into the regular mapping system.

In my case it is a small neighborhood park in an HOA which is accessible from three separate directions, by a walking trail only. All three walking trails are presently connected to the roads in the area. I originally set these up as Two-Way walking trails. The next day, I tested the route on my Waze app. To my surprise, Waze actually routed me through the park, which I cannot do in a car. https://www.waze.com/editor/?zoom=6&lat ... 24&env=usa

Today I changed each of the three trails to an unknown type and will check again later once the changes go live.

~Doug

[DwarfLord]

In the non-drivable roads section of the Road Types (USA) wiki, the figure that used to show how walking trails distort routing even when unconnected appears to have been replaced by another figure showing a tiny part of a map somewhere?

[EDIT: The previous figure is back. The above paragraph can be disregarded]

But more to the point, I have been testing that walking-trail routing distortion effect and cannot reproduce it.

If I understand the routing distortion, if a walking trail gets closer to the destination than anything else, one will be routed as close to the walking trail as possible and NOT as close to the destination as possible even if the walking trail is disconnected from the road network. However, while testing in a local situation I could not reproduce this.

The situation involves a parking lot followed by a 5-minute walking path leading to a theme park ("Roaring Camp") railway station. The Google pin is at the railway station, which is generally accepted as the heart of the theme park, not the parking lot some distance away. I have not tried to move the Google pin nor do I think it should be moved in this case.

Sharing a boundary with the theme park is a State Park with good roads. Those roads come much closer to the railway station than the Roaring Camp parking lot road.

• With no walking trail, routing to Roaring Camp takes one into the adjacent State Park.
  
• With a connected walking trail, routing to Roaring Camp takes one to the correct parking lot, but then apparently (according to a UR) one is advised one still has 5 minutes to go. This is in fact quite accurate but it surprised the driver who thought Waze wouldn't and shouldn't think about that.
  
• But if I disconnect the walking trail from the Roaring Camp parking-lot road, leaving a gap of just 10 feet or so, routing again takes the driver to the adjacent State Park.

So, I am leaving the walking trail connected for now because that's how to make routing to Roaring Camp work (without asking Google to move the pin, which I'm not sure I could persuade them to do and don't really want them to do, I like it where it is). Maybe that makes me a WME devil...

[DwarfLord]

Thanks so much for gathering and posting the references to the disparate threads. It would be great to consolidate this into a wiki entry except everything appears so hopelessly inconclusive. I was not aware that the routing distortion discussed in the wiki is not dependable or even consistently reproducible in the same setting. And here I thought I could at least take advantage of that behavior. Yikes.

some failures to get any route still seem to occur when the route is near a long walking trail. If I try to route to 1820 Wiehle Ave, Reston, VA, I can't get a route in the live map as discussed here. But the client now routes me to a point on Wiehle Ave.

I see what you mean. But if I try from 1820 Wiehle Ave to 1820 Wiehle Ave I do get a route -- Live Map says it will be 4.11 miles on the Washington & Old Dominion Bike Trail, and those 4.11 miles will take me 0 minutes. That's some fast biking.

[DwarfLord]

As far as I know, our guidance still says that Walking Trails that do not somehow improve routing for drivers are strongly discouraged.

But, Waze has that nice "Walking Trail" street type, and we have lots, and lots, and lots of new editors who don't spend much time with the Wiki or forums and are looking for something to contribute. I don't know about you all, but Walking Trails have been proliferating in my area and honestly I don't want to spend the energy involved in attempting to reach each and every one of their authors via PM. And then, quite possibly since so few of these folks check PMs, deleting a lot of someone's work. I hate deleting somebody's hard work just as much as they will probably hate seeing me do it.

But, walking trails that remain beget more walking trails, and more and more, as new editors see the existing examples and add their own. It's springtime for walking trails around here.

It would be different if these Walking Trails were obviously breaking things. But I have not been seeing URs pop up around them the way I would have expected if they really were so dangerous. And by current understanding they should be dangerous, as many are urban bike paths that come close to a multitude of destinations.

What do I do? Put on my walking-trail grinch hat and start PMing, answering questions, waiting, and eventually deleting, and find a good Waze therapist to help me through the guilt? Or limit the effort to disconnecting them where necessary, and accept that pretty soon we'll be covered in walking trails?

[DwarfLord]

I would leave them alone unless they are actually causing problems. My guess is they're acting like railroads act now, and are essentially invisible to the routing server. My opinion is that they look quite nice

I verified today that Waze will still route vehicles over connected walking paths. So if they are connected they are definitely not invisible; in my test, Waze routed me from a street over a connected walking path and thence onto a connected parking-lot road.

How they behave if completely unconnected, I haven't checked. If they were as bad as Alan's tests from a few months ago suggested, I would have thought we'd have URs everywhere by now. So I'm lost as to just how dangerous the disconnected ones are.

[DwarfLord]

Waze treats Walking Trail like a PLot or Private Dr in that it will avoid them as through routes, but will route onto them for destination or use them as a starting point.

Hmmm...given that, in my test, Waze routed me from a well-connected Street to a Walking Trail to a Parking Lot Road that was also well-connected -- meaning the transition penalty for WT to PLR was negligible -- do you suppose that a Walking Trail is a Parking Lot Road except with different rendering? That would explain a lot...maybe...I doubt Walking Trails register slowdown detections...hmmm...