[NEW] Best Continuation algorithm has been changed

The place to get information and ask questions about everything to do with properly and successfully editing the Waze Map.

Use this forum for all general editing questions, and the sub-forums for specific types of Waze Map Editor features.

Moderators: Unholy, bextein

Re: [NEW] Best Continuation algorithm has been changed

Postby PesachZ » Wed Sep 24, 2014 2:14 am

Taco909 wrote:Okay... back to example A just to confirm....
Being a mH with less than 45 degrees, it will instruct "exit right" when the side road is a "street"
Is a "stay" instruction is desired, the appropriate fix would be to add a 5m stub of mH onto the street and leave it unnamed, thus forcing BC to stick to the left, and the instruction to be "Stay to the right to Main St"
Correct?

I have an "exit" that I need to fix.. I overlooked mH generating exits (and honestly, getting exits on most MH in my area is annoying)

Alternatively changing it to a "turn" instruction also eliminates the "exit".

Sent using Tapatalk for Android 4.4.2
PesachZ
Wiki Master
Wiki Master
 
Posts: 4512
Joined: Mon Jul 01, 2013 12:51 am
Location: NY, USA (also NJ sometimes) {GC}
Has thanked: 1998 times
Been thanked: 2374 times

Re: [NEW] Best Continuation algorithm has been changed

Postby PesachZ » Wed Sep 24, 2014 3:22 pm

Taco909 wrote:
PesachZ wrote:Alternatively changing it to a "turn" instruction also eliminates the "exit".

Agreed... and in the case I was working on, perhaps "turn" would not add confusion, but we have a LOT of freeway onramps (and regular junctions) where "Stay to the right" is appropriate, but they are MH/mH roads according to California FC.

https://www.waze.com/editor/?env=usa&lo ... 605&zoom=7
Stay would perhaps be more appropriate for the upper ramp with the right lane being turn-only (I see it is set to "turn").
Stay could be questioned on the lower ramp, but it really is a shallow angle and looks more natural from the street than it does from the aerial.

IMHO, if there are traffic signals and driveways on the road that I'm on, I should not be getting an "exit" instruction. This is more common than not in heavily urbanized and suburban areas.

https://www.waze.com/editor/?env=usa&lo ... 605&zoom=6
Both upper ramps are going to generate "Exit"... This is a numbered California State Highway... with sidewalks and bicycles and driveways (my dentist is on this street) on-street parking, etc.... All of the indications of being "Main St USA" with a 40m median that is used as a series of local parks, and they even run a classic car show on the median.
The road by no means deserves an "exit" instruction (if for no other reason than there are no exit signs ;) )

It would be very nice if we had a way to specify the instruction (at least on MH/mH, Freeway is pretty absolute for "exit") without having to fudge a wayfinder.

I don't see an easy way to do it automatically. In my area, it would be easy to say "If the turn is less than 45 and the outgoing segment is not "ramp" then use "Stay"... but that's not going to work for rural highways that are posted with "exit" signs.

Now this guy here:
https://www.waze.com/editor/?env=usa&lo ... ts=4717076
BC would be to the left, but the name changes.
Mismatched type is to the right, but the name remains the same.
Is this going to produce "Exit right to Turnbull Canyon"?

turnbull.jpg



And maybe this is a better topic for a split-off thread.

I don't want to go too far off topic, but I don't think it's a good idea to start making wayfinder stubs just to change an instruction from exit to stay. They will be ankle to be controlled when the junction box aka big junction becomes available.

Sent using Tapatalk for Android 4.4.2
PesachZ
Wiki Master
Wiki Master
 
Posts: 4512
Joined: Mon Jul 01, 2013 12:51 am
Location: NY, USA (also NJ sometimes) {GC}
Has thanked: 1998 times
Been thanked: 2374 times

Re: [NEW] Best Continuation algorithm has been changed

Postby PesachZ » Sun Oct 19, 2014 1:07 am

shawndoc wrote:


The left fork is at 1, the right fork is at 45. So Waze will only give a right turn command, the left is seen as a continuation. Both options need to be under 45, and as close to each other as possible in degrees to give a keep right/left. Also, you may need to add a street name to the left fork.


shawndoc wrote:


One angle is 48, the other is 37. Try reducing them both to under 45, and try to get the angles as close to each other as possible. The way it is set up now, the Left fork is considered a turn and the right fork is considered a continuation.

As long as both angles are less than 45° (preferably for safety aim for less than 35° to avoid accidentally crossing the threshold), it doesn't matter how close they are to each other.

One side can be 40° to the left and the other 40° to the right, with a total of 80° in between them, and it would still work.

Where are you seeing that they need to be collude to each other, if that info is our there is like to correct it.

On a side note, due to the issue caused by overlapping segments at a junction node, and the inadequacies of the closure reporting system in the client; I aim to have at least 20°- 30° between adjacent segments at a split or exit.

Sent using Tapatalk for Android 4.4.2
PesachZ
Wiki Master
Wiki Master
 
Posts: 4512
Joined: Mon Jul 01, 2013 12:51 am
Location: NY, USA (also NJ sometimes) {GC}
Has thanked: 1998 times
Been thanked: 2374 times

Re: [NEW] Best Continuation algorithm has been changed

Postby PesachZ » Sun Oct 19, 2014 5:13 pm

shawndoc wrote:
t0cableguy wrote:Level 3 again these are locked to 4... bet you guys are sick of me lol
https://www.waze.com/editor/?env=usa&lo ... 2,66126422


I would suggest PMing TXEMT or Banished since they are last to touch the junction to find out why it is set up the way it is. Waze doesn't handle 3 way splits, and that's not how you would normally set that up. No matter what you're not going to get ideal instructions there.

Waze *does* (kinda) handle 3 way splits BTW, just not necessarily the way you'd expect. The middle branch always gets a right sided instruction regardless of angle. A PM to the last editor would still be a good idea to see if it is still functioning as intended.

EDIT: This is not to be misconstrued as guidance that creating 3 way splits are OK, or encouraged. Just sometimes there are extraneous circumstances which are corrected with extreme edits. When it was done by a champ, I err on the side of ask first, fix if needed later.
Sent using Tapatalk for Android 4.4.2
PesachZ
Wiki Master
Wiki Master
 
Posts: 4512
Joined: Mon Jul 01, 2013 12:51 am
Location: NY, USA (also NJ sometimes) {GC}
Has thanked: 1998 times
Been thanked: 2374 times

Re: [NEW] Best Continuation algorithm has been changed

Postby PesachZ » Mon Oct 27, 2014 2:21 am

pumrum wrote:I know it has been discussed with WazeHQ, and perhaps they have even made a final determination in the secret forums -- but can we get the alternate naming on ramps issue logged as an official bug on the bug sheet? I think most of us agree that to have a properly configured ramp which branches from a freeway/highway not give an exit instruction is bad. at least this way they can see that it really is causing us issues (I get reports on this every day which I have to go out and fix) and maybe even consider coding a simple exception to their algorithm for ramps. :?

I've been told they said this change was an intended of enhancing their algorithm to better be able to determine actual continuations. That being said it would have been nice to get a heads up before hand.

They may not actually have anticipated the results to our wayfinder system. We designed the wayfinder as a workaround to their algorithm, they may not have considered the effects on it.

Now, I don't expect any changes in the best future, as their priorities are set on getting the junction box ready for roll out. That will hopefully give us more and better options, eliminating much of these issues in its wake. For now this is what we have, and based on what I've heard 3rd hand from Waze, don't expect any changes in the immediate future.

don't shoot the messenger

Sent using Tapatalk for Android 4.4.2
PesachZ
Wiki Master
Wiki Master
 
Posts: 4512
Joined: Mon Jul 01, 2013 12:51 am
Location: NY, USA (also NJ sometimes) {GC}
Has thanked: 1998 times
Been thanked: 2374 times

Re: [NEW] Best Continuation algorithm has been changed

Postby PesachZ » Mon Oct 27, 2014 2:58 am

pumrum wrote:
PesachZ wrote:They may not actually have anticipated the results to our wayfinder system.


I wouldn't care half as much if it was just wayfinders, but this is affecting properly configured exits too. Instances where you have a continuous freeway with one ramp segment that branches out to the right, and because the alternate names match on the input freeway and the output ramp, you don't get an "Exit Right" instruction. This happens every time a US or State highway splits from an interstate, for example. This is the most obvious situation I run into in CT, but I have to imagine there are tons of others (wayfinders included) that will continue to haunt us all over the developed map. Ramps are intended to be a protected class of segment. If they weren't, we'd just make all offramps freeways (yes, i know they make the client zoom in and hide the text, but those are just interface features, and it was really nice to have the functional features too).

Can't blame you for being the messenger :) - very much appreciate the input. i still do think we should add it to the official bug sheet. I would add it, but I haven't been able to request modify access to the sheet

These changes didn't truly change the function of ramps, even before these changes they weren't 'protected'. There has been a push to force an instruction from any (highway/freeway) segment to a ramp, but that has not happened yet. It is still on the wish list. Before these changes the functional differences of ramps were limited to handling elevation changes, preventing routing transition penalties, and the GUI features you mentioned.

A ramp which had a matching primary name to a freeway for example, where the freeway continuation had a different name, would not have got an instruction in the previous algorithm either.

The only thing which has changed recently is what's considered matching. How a match is handled, and how a non-match is still the same.

When presented to Waze that this is disastrous and must be fixed by Alanoftheberg, he was told that this is not a bug. Addinng it to the bug list would probably be of no consequence. The bug list is to bring things to Wazes' attention, they've already been made aware of, given links to the forum thread discussing it, and responded. Adding it to the bug list, won't tell them anything they don't already know, and it will likely be ignored there.



Sent using Tapatalk for Android 4.4.2
PesachZ
Wiki Master
Wiki Master
 
Posts: 4512
Joined: Mon Jul 01, 2013 12:51 am
Location: NY, USA (also NJ sometimes) {GC}
Has thanked: 1998 times
Been thanked: 2374 times

Re: [NEW] Best Continuation algorithm has been changed

Postby PesachZ » Tue Dec 09, 2014 7:02 am

taco909 wrote:Bumping this one for a new question.
In light of the revised BC, is there any functional difference between these two junctions?

bctest.jpg


Every now and then I'll work a UR where someone did not get a "turn" instruction at the end of a block.
I always assumed it was BC squelching the instruction since there is only one way to go, but in some cases, such as a "square" block with a single entrance from the feeder, or worse when the roads zig-zag as is not uncommon in the hills, it can be confusing with no turn instructions once inside.

I recently considered that the lack of instruction MAY be the lack of a 46+ degree junction angle.
The app doesn't resolve to this detail anyways, so removal of the extra geonodes is of course a good thing... but will "cleaning up" these corners in neighborhoods result in a change in turn instructions without a "stub" to nowhere?

(I only used two geonodes to smooth the transition for simplicity, in actuality on the map, 4 and sometimes 6-8 are used to make it a smooth curve so the angle at the actual junction node is frequently nearly 180)

both of your examples will NOT have any instruction. Because the junction only cosists of 2 segments, there is no other option restricted or otherwise period, and therefore Waze considers this the ultimate BC. If there are ONLY 2 segments connected at a junction, there will be no instruction ever, regardless of name, type, angle, etc.. However this does not apply if there more than to segments connected, but only one of them has a green arrow.
This is not new, this is by design to prevent unnecessary instructions, and has always been the case. See step 1 in the Algorithms list of conditions.
PesachZ
Wiki Master
Wiki Master
 
Posts: 4512
Joined: Mon Jul 01, 2013 12:51 am
Location: NY, USA (also NJ sometimes) {GC}
Has thanked: 1998 times
Been thanked: 2374 times

Re: [NEW] Best Continuation algorithm has been changed

Postby PesachZ » Wed Dec 10, 2014 6:59 am

t0cableguy wrote:unless things have changed with these i have had turn rights and lefts instructed in the past when they were set at right angles. curving them into straights junctions has supressed the instructions for me.

These may be cases where those other roads therefore deserve to mapped (as private or PLR, trail, or something, for the value they add to the regularnroads navigation instructions. Review that with your local SM or RC.
taco909 wrote:
t0cableguy wrote:unless things have changed with these i have had turn rights and lefts instructed in the past when they were set at right angles. curving them into straights junctions has supressed the instructions for me.

I would of course expect suppression if the junction is less than 45 degrees.

I'm not surprised at the lack of instruction on the 90s, as I have worked URs on that, but IMHO, it can be very confusing, where you make your turn, and Waze says "In 2.5 miles, turn left on West St",
but in the next 2.5 miles, you have three 90 degree turns onto different named roads.
In some cases, these junctions MAY be at an unmapped driveway or dirt farm road. Unless the driver looks at the app, they may miss the turn.

I'd love to see an example of a junction of any angle, with ONLY two connected segments, giving a turn instruction. Please provide a PL.

Sent using Tapatalk for Android 4.4.2
PesachZ
Wiki Master
Wiki Master
 
Posts: 4512
Joined: Mon Jul 01, 2013 12:51 am
Location: NY, USA (also NJ sometimes) {GC}
Has thanked: 1998 times
Been thanked: 2374 times

Re: [NEW] Best Continuation algorithm has been changed

Postby PesachZ » Tue Apr 28, 2015 7:42 am

I wanted to warm everyone with a quick heads up, we found some bugs in the BC algorithm, and Waze is about to fix it. This may have some effect on existing roads but hopefully only by adding instructions where there were none before. In some cases where it is now considered BC, they will start to get an instruction. I hope to have more details soon. Again this change only effects roads with some alt names.

Sent using Tapatalk for Android 4.4.2
PesachZ
Wiki Master
Wiki Master
 
Posts: 4512
Joined: Mon Jul 01, 2013 12:51 am
Location: NY, USA (also NJ sometimes) {GC}
Has thanked: 1998 times
Been thanked: 2374 times

Re: [NEW] Best Continuation algorithm has been changed

Postby PesachZ » Thu Apr 30, 2015 10:46 am

qwaletee wrote:Thanks PZ, any ETA on more info?

I expect about two weeks.

Sent using Tapatalk for Android 4.4.2
PesachZ
Wiki Master
Wiki Master
 
Posts: 4512
Joined: Mon Jul 01, 2013 12:51 am
Location: NY, USA (also NJ sometimes) {GC}
Has thanked: 1998 times
Been thanked: 2374 times

PreviousNext

Return to Waze Map Editor

Who is online

Users browsing this forum: No registered users