User:Voludu2/WT notes View history

Revision as of 11:10, 21 May 2015 by Voludu2 (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

TheLastTaterTot leads discussion on recent experiments in exotic applications of Walking Trails:

Round Table 1

Patti Tatertot
This is what Otto said: "Here is the conclusion we reached the other night in New York. If you want people to be routed to the end of a WT, then connect it, but disallow the inward turns, just in case. If the destination pin is closest to that WT segment, the routing prompts and purple line will end where the WT starts, but the destination flag will be visible at the actual final point."
I do not believe that is 100% true
Patti • Sun, 4:57 AM

Pesach Z
🔒
Pesach • Sun, 4:58 AM


Patti Tatertot
Thanks
Patti • Sun, 4:58 AM

Pesach Z
What do you believe is not true about that?
Pesach • Sun, 4:58 AM


Patti Tatertot
though I do not know what he means by "end of the WT"
he's currently in the middle of a couple other conversations, I think.. So I haven't gotten a response from him.
Could you tell me what you guys mean by that?
Patti • Sun, 4:59 AM


Pesach Z
OK let me find an example
Pesach • Sun, 4:59 AM


Patti Tatertot
okay
Patti • Sun, 5:00 AM


Pesach Z
https://www.waze.com/editor/?env=usa&lon=-73.85801&lat=40.81385&zoom=7&segments=81523089
lets look at a route to 7 Murray Ct, Bronx NY
https://www.waze.com/livemap?lon=-73.85791&lat=40.81405&zoom=15&from_lat=40.81771&from_lon=-73.85748&to_lat=40.81405&to_lon=-73.85791
Pesach • Sun, 5:01 AM


Patti Tatertot
So by "inward turns", is he referring to the turns from the WT?
not turns onto the WT
Patti • Sun, 5:02 AM


Pesach Z
No, kets just look here for a min
Pesach • Sun, 5:02 AM


Patti Tatertot
k
yes, this is the kind of behavior I would expect from a WT
Patti • Sun, 5:02 AM


Pesach Z
a week ago this segmetn was close to but not connected to soundview.
Pesach • Sun, 5:03 AM


Patti Tatertot
That's connected with turns allowed onto it.
Patti • Sun, 5:03 AM


Pesach Z
it was closer to soundview than bolton, but a route to 7 Murray CT, still ended on bolton near the segment
Pesach • Sun, 5:03 AM


Patti Tatertot
YES
Patti • Sun, 5:03 AM


Pesach Z
when connected it forces the route to go to the end of the WT that is connected to the network (in this cae the end of the WT is the node at soundview)
BUT if you look in livemap the purple line does turn onto Murray (the WT). the red arrows leaving the WT prevent any through routing or uturns.
Pesach • Sun, 5:05 AM


Patti Tatertot
so by "end of the WT", you guys mean the end that connects to the drivable segment?
Patti • Sun, 5:05 AM


Pesach Z
If you want to prevent the pruple line from making tht last trun onto the WT, but still route to proper end of the WT (soundview) you connect it but dont allow the turns into it.
yes End=junction, the end we care about is an end junctioned where the WT can be accessed by drivers
refresh wme and look at the turns now
Pesach • Sun, 5:07 AM


Patti Tatertot
okay
let me show you some screenshots of my WT experiment in Providence
Patti • Sun, 5:07 AM


Pesach Z
this is how he was saying to do it so you dont get the last bend of the purple line, But the flag will still show up in the same place, and route will take you through soundview and not bolton
Pesach • Sun, 5:08 AM

Patti Tatertot
I haven't finished my experiment in Queens bc I don't have access right now... But I didn't see a difference between disconnection vs overlapping with no junction.
okay, so I think there is another rule involved
And it's not as simple as that.
I'll show you examples.
Patti • Sun, 5:09 AM


Pesach Z
I thought you said you DID see a difference between just disconnected, and overlapping
Pesach • Sun, 5:09 AM


Patti Tatertot
Nope
Oh wait
Patti • Sun, 5:09 AM

Pesach Z
that when it was overlapping Waze would route to either end
Pesach • Sun, 5:09 AM


Patti Tatertot
let me restate so it's clear
Patti • Sun, 5:09 AM


Pesach Z
i'll brb, you type
Pesach • Sun, 5:09 AM


Patti Tatertot
k
I'm going to enumerate, so the points that I'm trying to make is more distinct.
So, there are 4 types of interactions with WT. (1) Disconnected and not overlapping, (2) Disconnected and overlapping, and (3) Connected, turns onto it disabled, (4) Connected, turns onto it enabled
For the 4th type where the WT is connected to a drivable road via a junction AND has turns onto it enabled, it behaves just as you would expect it would and routes people onto the WT segment to the destination.
(2) and (3) behave the same. That is, whether it is disconnected+overlapping or connected+turns onto it disabled, routing to a destination closest to the WT is the same.
The first type also behaves like (2) and (3), but that gets slightly tricky bc it's harder to measure distances to nearby roads, so harder to predict the behavior... unless the difference in distance is obvious.


I argue with the statement about type 1
Pesach • Sun, 5:17 AM


Patti Tatertot
Sure, I haven't experimented with the first one as much.
I'd really like to hear about what you've found
Patti • Sun, 5:17 AM

Pesach Z
An address on the WT, closest to the WT, the WT was closest to Soundview but the route ended on Bolton
Pesach • Sun, 5:17 AM

Patti Tatertot
But first let me explain what i've found
😛
Patti • Sun, 5:17 AM

Pesach Z
I thought you were done
I'll be quiet
Pesach • Sun, 5:19 AM


Patti Tatertot
What did you get for when the WT is connected, but with turns ONTO it disabled? (type 3)
Patti • Sun, 5:19 AM
Pesach Z


IIRC the flag is at pin location, the purple line stops at the red arrow
Pesach • Sun, 5:21 AM


Patti Tatertot
Btw, for the sake of simplicity, I'm going to lump type 2 and 3 together. More testing needs to be done, but let's just assume they are the same for now.
Did you test several start origins?
Patti • Sun, 5:23 AM


Patti Tatertot
anyway
Here is what I get for a connected WT with turns onto it disabled...

And looking at my screenshots that I took of the WT disconnected+overlap, I misspoke about type 2 and 3 being exactly the same. The behavior was exactly the same for the Queens WT experiment, probably because of its configuration. The two types of WT interactions produce slightly different behavior with the road configuration here in my Providence WT experiment. But I'll save that point for another time, since I first want to convince you that the WT behavior for connected+no turns (type 3) is actually more similar than you may think to type 1 – disconnected+not touching.
Anyway, I need a brief nap. Been up too long. ☺ See you in a few hours. Here is the PL to the WT in Providence:
https://www.waze.com/editor/?env=usa&lon=-71.40141&lat=41.82326&zoom=7&segments=85272560 It's currently connected with turns allowed.
Patti • Sun, 5:53 AM

Given your screenshot I believe you,
So we're back to allow turns in but not out
Pesach • Sun, 6:01 AM


Otto Plunkett
Ugh. This is a problem, because we really need an option the takes us TO the appropriate end of the WT segment, but not onto it. If we allow the turns, the purple route line and prompts will at least sometimes instruct the driver to turn onto an undrivable road.
Otto • Sun, 1:45 PM


Patti Tatertot
i think it's possible
But need to combine with PB
THere's a rule to this..
Patti • Sun, 1:46 PM


Pesach Z
Pb isn't routable
Pesach • Sun, 1:46 PM


Patti Tatertot
But I hesitate saying it now
Yes, that's why you combine with PB
welll I guess you don't have to have PB
you could also make it completely detached to any road... but you would need to check the distances between all streets
I dunno.. I hesitate to say any more bc I need to run some tests
Patti • Sun, 1:48 PM


Pesach Z
Disconnected doesn't work, we proved that
Pesach • Sun, 1:48 PM


Patti Tatertot
it does
In some cases
Patti • Sun, 1:48 PM


Pesach Z
We need all cases
Pesach • Sun, 1:48 PM

Patti Tatertot
Right - the question is why does it work in some cases
Rather than simply saying that it doesn't work, let's toss that idea out, I think it's
important to find out why it's not working in some cases.
Patti • Sun, 1:49 PM


Pesach Z
Any ideas?
Pesach • Sun, 1:51 PM


Patti Tatertot
Hmm I have some theories
but none that I am certain about
Patti • Sun, 1:52 PM


Pesach Z
Wanna share? We can try to pool a our knowledge
Pesach • Sun, 1:53 PM


Patti Tatertot
Yeah, should we create another GHO?
Patti • Sun, 1:54 PM


Pesach Z
Sure if you'd like
Pesach • Sun, 1:54 PM


Patti Tatertot
we can stay here too though. I just wasn't sure if everyone wants to get pinged about WT talk. lol
Patti • Sun, 1:55 PM

Lol ok ok -- i wasn't sure if people were interested in WT
so I think that Waze mgiht be routing to the end nodes of WT
So the idea is... If you made a U, such that both ends are connected to the same street But turns disallowed
It will work
This isn't necessarily a solution
but if this were true, it gives us insight into its behavior
Patti • Sun, 2:02 PM

Pesach Z
Are you thinking it is routing to the segments closest to the end node closer to the pin
Pesach • Sun, 2:02 PM

Patti Tatertot
Nope, it is routing to the end node closest to you
One sec
I'll show you an example with the numerous WT in CA
So these are WT in CA that are disconnected
I think the behavior is more consistent when connected, tho I don't know the exact reason yet. Let me see if I can find the PL
https://www.waze.com/livemap?zoom=16&lat=37.11649&lon=-121.64837
connected+no turns, that is
https://www.waze.com/livemap?zoom=17&lat=37.11163&lon=-121.64211
not the best example
saw a better example three days ago, but I saved the wrong PL
Patti • Sun, 2:14 PM


Otto Plunkett
Question: Patti, when you say "overlapping", do you mean actually one portion of a segment lying partially on top of the other, in parallel (our chosen definition of overlap in WME parlance), or merely crossing?
Otto • Sun, 2:16 PM


Patti Tatertot
sorry
yes crossing
Patti • Sun, 2:16 PM


Otto Plunkett
Ok
Just wanted to be clear.
Otto • Sun, 2:16 PM


Pesach Z
crossing at the same elevation
Pesach • Sun, 2:16 PM


Otto Plunkett
Right
Otto • Sun, 2:16 PM


Patti Tatertot
elevation doesn't matter
They tend to set these WT to -5 in CA
https://www.waze.com/livemap?zoom=16&lat=37.11855&lon=-121.64569
Better example
Even better
https://www.waze.com/livemap?zoom=17&lat=37.12052&lon=-121.65336
Patti • Sun, 2:19 PM


Pesach Z
I gtg ttyl, It'd be interesting to find a theory that always works
Pesach • Sun, 2:20 PM


Patti Tatertot
k
ttyl
Hmm theory revision --- the algorithm might be balancing several factors: (1) distance between the end of the drivable road to the WT, (2) distance of the walk, and (3) distance driven to get to that point
https://www.waze.com/livemap?zoom=17&lat=37.11999&lon=-121.64674
Patti • Sun, 2:27 PM


Otto Plunkett
For what it's worth, the -5 elevation was a legacy thing, dating back to when matching elevations made if more likely for Waze to route onto something undrivable, like a railroad.
Otto • Sun, 2:28 PM


Patti Tatertot
yeah that's what I thought
Patti • Sun, 2:28 PM


Otto Plunkett
(it's not current policy)
Otto • Sun, 2:28 PM


Patti Tatertot
This WT in CA needs to go.. Or needs to be connected to the road system.
or converted to PB
Patti • Sun, 2:29 PM


Otto Plunkett
I will look at it in a minute.
Are there addresses on it?
Otto • Sun, 2:36 PM


Patti Tatertot
yes
Patti • Sun, 2:36 PM


Pesach Z
I'll bet that this is an undocumented balancing act, and that even the devs couldn;t tell you exactly how it works
it's not something they planned for
Pesach • Sun, 2:39 PM


Otto Plunkett
As evidenced by their expressed intent with WT at the meetup.
I think you're right.
Otto • Sun, 2:39 PM


Pesach Z
well the same rules should apply to any non connected driveable segment
since right now WT is just another driveable segment with a penalty
Pesach • Sun, 2:40 PM


Otto Plunkett
Good point.
Otto • Sun, 2:41 PM


Patti Tatertot
Hmm very interesting
that is a really good point re rules to any nonconnected segment
nonconnected, routable segment
Patti • Sun, 2:48 PM


Dovid Gross
What plans did you hear about WT?
Dovid Gross • Sun, 5:18 PM


Otto Plunkett
Waze didn't understand the usefulness of having a non-drivable segment type which was in fact routable. They expressed the possible intention of making WT more like Pedestrian Boardwalk, appearance only.
Otto • Sun, 5:35 PM


Dovid Gross
That's break some stuff. Do they now understand the usefulness of allowing a route to go beyond the drivable portion?
Dovid Gross • Sun, 5:48 PM


Otto Plunkett
It sparked a rather intense interchange. I think we have them convince to leave things alone for the time being, such that we can work with them to decide the best fix(es).
... or they will just change something and surprise us, as it SOP.
😀
Otto • Sun, 5:50 PM


Pesach Z
We showed them.how we use, they were unaware. Apparently other countries (Israel) don't use the same way
Pesach • Sun, 5:51 PM


Otto Plunkett
Everyone agrees that Walking Trail is an inappropriate name for how the segments work.
Yeah, what Pesach said. They were completely unaware of our use case.
Otto • Sun, 5:51 PM


Patti Tatertot
How does Israel use WTs?
Patti • Sun, 6:04 PM


Pesach Z
For … wait for it … walking trails, that have no routing need
Pesach • Sun, 6:05 PM


Patti Tatertot
lol

  • gasp*

Patti • Sun, 6:05 PM


Otto Plunkett
Our big problem here is that we also have tons of actual walking trails typed that way, so however they fix the problem, merely renaming the road type, creating a new one for us, or whatever, we will have a big job to do, changing all of ours that are not what we then want them to be.
Otto • Sun, 6:09 PM

Pesach Z
We almost definitely have a big job ahead of us but hopefully will have advanced warning and a plan to do it right
Pesach • Sun, 6:10 PM


Otto Plunkett
Some people hope for world peace, too.
Sorry for being snarky.
Otto • Sun, 6:11 PM

Pesach Z
Just think what the world would be like without them
Pesach • Sun, 6:11 PM

Otto Plunkett
I'm just so afraid that a big switch is gonna flip.
Otto • Sun, 6:11 PM

Pesach Z
if this is the compromise they get what would the compromise be if they weren't even asking If we didn't say anything that would have been a big flip regardless at least you have a chance now
Pesach • Sun, 6:12 PM


Otto Plunkett
I'm happy with any concession from Waze.
Yep
So we need to come up with a couple contingency plans.
Otto • Sun, 6:13 PM


Dovid Gross
I would vote for them to convert all existing walking trails to the new type
At worst, there will be some real walking trails that we want to switch back, but it shoudln't hurt anything
Dovid Gross • Sun, 6:14 PM

Pesach Z
And then we can change the bad ones back to the no. Routable type.
Pesach • Sun, 6:14 PM


Dovid Gross
if they won't do that for us, then I guess we can ask for a report of every segment currently types walking trail so we can review and fix ourselves
Dovid Gross • Sun, 6:15 PM

Pesach Z
Exactly, but they may not want to create a new type, so there may have be a juggle
Pesach • Sun, 6:15 PM

Dovid Gross
exactly, PZ
Dovid Gross • Sun, 6:15 PM

Patti Tatertot
Yes, rename WTs and PB. PB perhaps can be the new unroutable WT
Patti • Sun, 6:15 PM

Dovid Gross
well, either they are cretaing a new type OR adapting to our usage OR throwing us under th ebus with no option
Dovid Gross • Sun, 6:15 PM

Otto Plunkett
I think we made our case pretty well.
Otto • Sun, 6:16 PM

Pesach Z
Pedestrian (routable) and (unroutable). They can even then prevent the purple line from showing on it.
Pesach • Sun, 6:16 PM

Otto Plunkett
😀
Otto • Sun, 6:16 PM

Pesach Z
They said they need to discuss
Pesach • Sun, 6:17 PM

Patti Tatertot
I'm guessing that this has become a topic because other countries are having problems with people mapping WT and having those affect routing?
Patti • Sun, 6:17 PM

Pesach Z
Correct most likely, and because there's been a disagreement in the form on how we should use them
Pesach • Sun, 6:20 PM

Patti Tatertot
oh man.. so there's three options right now?... we get shafted, or the rest of the world deals with the routing problems, or Waze spends time and effort to make some changes in the road types.
Patti • Sun, 6:25 PM


Pesach Z
pretty much patti
one thing we do know, there are no plans for walking or cycling navigation in the futur Pesach • Sun, 6:27 PM

Dovid Gross
whole 'nother set of restrictions they'd have to have, and vehicle choice type, and figuring out speed data pollution from one type to another... yeah, outside the core mission and too much trouble to bring it in
Dovid Gross • Sun, 6:29 PM

Craig M
So PB isn't routable, meaning it won't give instructions to follow that road. Waze has also said to not map trails. So why leave that as a mapping option?
Craig • Sun, 6:30 PM

Dovid Gross
color
visual interest
Dovid Gross • Sun, 6:31 PM

Patti Tatertot
also prevents MPs when people use Waze incorrectly... such as for biking
Patti • Sun, 6:32 PM

Pesach Z
They exist for the cases where they are useful as a reference to drivers. (A pb crosses a road with junction, out an overpass bridge used as a spatial reference)
Pesach • Sun, 7:42 PM


Otto Plunkett Also, there are things which can be mistaken for roads when viewed in the aerials. If we map them correctly (as PB or whatever) and lock them, then uninformed editors won't come along and create routable roads there over and over.
Otto • Sun, 8:48 PM

Pesach Z
^^
Pesach • Sun, 8:49 PM

Round Table 2 18 May 2015

Hey Pz, no problem
I figured it out and it works
I made microdog legs and overlapped the segment completely
and just checked the tile update and BC works
Patti • 9:18 AM


Pesach Z
Were you trying to eliminate any instruction at the those two segments?
Pesach • 9:18 AM


Patti Tatertot
yup
Patti • 59 mins


Pesach Z
Where?
Pesach • 59 mins


Patti Tatertot
Oh, it was just an experiment.. I don't like it, but it works.
One sec.
https://www.waze.com/editor/?env=usa&lon=-152.33418&lat=69.64945&zoom=8&segments=85770628,85770630
Patti • 59 mins


Pesach Z
Went are they separated at the west end like that?
Pesach • 56 mins


Patti Tatertot
west end?
went?
Patti • 56 mins


Pesach Z
Why are …
Pesach • 55 mins


Patti Tatertot
oh, so that they give no instruction going in from left or right
See -- don't like it.
I like how it doesn't tell them to turn onto the WT but I don't like how it doesn't tell them that they have reached their destination And gee thanks, Pz for not telling me that these WTs behave like Dirt road 😛
Had to figure htat out myself
lol
Patti • 54 mins


Pesach Z
Oic you were trying to eliminate a turn instruction, not a keep
Pesach • 54 mins


Patti Tatertot
Yup
Patti • 53 mins


Pesach Z
Its in the forum😜
Pesach • 53 mins


Patti Tatertot
lol yeah I saw that, but a little too late.
Pbbthh
lol
I'm just kidding ☺
Patti • 53 mins


Pesach Z
To be honest we only figured it out about a month ago
Pesach • 53 mins


Patti Tatertot
I kinda suspected for awhile
livemap behaved unpredictably sometimes..
but my Waze app was consistently going over the WT
Whereas Otto's was not...
Patti • 52 mins


Pesach Z
Ah
Pesach • 51 mins


Patti Tatertot
Only difference was that I have short dirt roads enabled
Turns out, the best choice for real WTs that are mapped is to junction all of them, but disable all turns.
It gives the best results -- that is, IF the WTs have to stay on the map...
I've read through the first 3 pages and the last 2 pages of the WT wiki talk on the forum. stlil need to look through the middle 4 pages.
I also think I've found a way to consistently route to the WT junction (with turns disabled) but it only works situationally
will be difficult to pull off with unwieldy WT paths Patti • 45 mins


Pesach Z
If its inconsistent then the best option is still to allow turns in, no?
Pesach • 42 mins


Patti Tatertot
Hmm not really
Patti • 29 mins


Patti Tatertot
Check out my play area https://www.waze.com/livemap?lon=-152.31795&lat=69.64885&zoom=15 I'll label them for you.. give me a sec..
Patti • 40 mins

Patti Tatertot

If you play around with them, you'll see that you'll have the greatest success with this bottom right one.
It's not perfect, but it's a whole lot better than the others. You could try it also in the app by navigating to Tatertots, AK
Patti • 29 mins


Pesach Z
How are you measuring "success"?
Pesach • 28 mins


Patti Tatertot
Here, it's not systematic
Patti • 26 mins


Patti Tatertot
This is just for play I've been doing more systematic tests, but have yet to move onto complicated WT designs yet. For basic linear, single stick WTs, I've figured out a rule that will get you to the junction 99% of the time a junction with turns disabled I have yet to generalize the rule to more complicated WTs or over multiple streets. multiple crossing streets^
Patti • 26 mins\

Pesach Z
What is your rule?
Pesach • 18 mins


Patti Tatertot

https://www.waze.com/livemap?lon=-152.29978&lat=69.6484&zoom=15&from_lat=69.6484&from_lon=-152.29978&to_lat=69.64778&to_lon=-152.30735 Works from any direction
livemap will sometimes not draw the line all the way to the junction, but if you look at the distances in the turn-by-turn instructions, it adds up to the distance to the junction.
Basically, the side that you don't want it to get routed to needs to be farther from a drivable segment than the side that you do want it to get routed to. The end doesn't matter it seems.
Patti • 12 mins


Pesach Z
But I've had that fail with the same setup, and rout to the wrong side even though there was a junction.
Pesach • 11 mins

Patti Tatertot
I can have that east end of the WT overlapping the street (but not junctioned) and it won't matter
Unfortunately, I had to remove it to test more configurations bc I was running out of room at the time.
Patti • 11 mins


Pesach Z
We had different results here in NYC if you remember
Pesach • 11 mins


Patti Tatertot
Yeah do you have the PL handy?
Patti • 10 mins

Pesach Z
I fixed it though
Pesach • 2 mins


Patti Tatertot
i just want to see the streets
neighboring streets Patti • 2 mins

Quick Question 20 May 2015

Patti Tatertot
Hey Pz, do you recall back when you had disabled inward turns, how you got your Bronx WTs to fail? I recreated a miniversion of it and it looks like it's routing just fine (for the most part) I can't seem to get it to misroute me, but there is a small segment of Underhill Ave that fails to route at all.
Patti • Wed, 9:41 PM

Pesach Z
It has seemed to work then I showed it off and it wasn't working just routing from nearby
Pesach • Wed, 9:49 PM


Patti Tatertot
Do you recall if it actually provided an incorrect route or did it fail to find a route?
Patti • Wed, 9:50 PM

Pesach Z
It routed to the street behind
Pesach • Wed, 10:21 PM

Patti Tatertot
ah thanks Weird. I can't get it to do that. hmm
Patti • Wed, 10:31 PM