Maybe there is discussion of this happening elsewhere but I see the new tab in the WME. Any guidance on lane guidance (pun intended!)
Gav…
Maybe there is discussion of this happening elsewhere but I see the new tab in the WME. Any guidance on lane guidance (pun intended!)
Gav…
For example here: https://www.waze.com/editor/?env=row&lon=152.93687&lat=-27.51052&zoom=8&segments=192567499
After just having a quick play would you have to remove the slip lane, bowtie the intersection and then add three lanes with the associated turn left (2 left lanes), right (right lane) and straight ahead (right lane) instructions to get all the lanes to work here?
Yes, this is being actively discussed in the Discord server.
Look forward to the summary
I don’t think there’s a real summary yet, it’s being tested in selected areas where those who have access to guidance in the app can drive over segments to verify things. If you can jump on the Discord server into the Lane Guidance channel, it’s quite comprehensive.
My take is:
It’s being tested, limited people can add guidance, it’s not in production yet. I think the main thing to note right now is not to adjust segment geometry too much if it takes away visual feedback for drivers that don’t have lane guidance access (I think I got that right).
Because how quickly things are changing in relation to this, the Discord server is the best place to participate as information is changing quickly and the forums are too slow to keep up with those changes until some kind of formal guideline is developed for Australia in how lane guidance is implemented (some way off at this stage I’m thinking).
OK so how does one know if they have lane guidance available in the app? Is it only the beta channel?
At the moment it’s only in the beta app yes.
OK this will be long with the explanations so here is the TL;DR (explanation in following post)
Thoughts?
Gav…
Possible change #1. A “lane” must be at least 50m long. This is because we have lots of little turning lanes in Australia that are full width but I think would be silly to map. Note this would be the default position and specific confusing intersections could be approved by the community. Here is an example of what I wouldn’t map followed by an example which is more than 50m long that I would map.
https://www.waze.com/editor/?env=row&lon=152.91602&lat=-27.53011&s=18626312191&zoom=9
https://www.waze.com/editor/?env=row&lon=152.93755&lat=-27.51028&s=18626312191&zoom=8
Here some pictures that I was thinking would be useful for us https://imgur.com/gpg6MK1 In the first pic - The green line shows 50m of length. The red line shows the end of the full width lane. The yellow circle shows the lane markings we have when the lane is not full width. The second pic shows where the 50m is taken from; either the stop line or the tip of the marked or solid island. It also shows a turn left anytime lane that would not be mapped (westbound to southbound) and one that would be mapped (northbound to westbound)
Possible change #2. When adding lane guidance which has straight ahead as an option use TLS override for CONTINUE. I have been using the second map link above to trial things. I have found that if you are coming from the south and want to continue straight on towards Almay St then you will get no guidance (lanes or otherwise) as you approach the intersection. Also GBSpeedy, in a PM, told me about an issue that the USA community identified with exits close together and getting into the turning lane one exit too early. I believe this would also be solved by a continue as you would receive a continue going past the first exit and then guidance onto the second exit. [[EDIT: GBSpeedy tells me a new feature is coming in the WME to force lane guidance indications for straight ahead guidance when no other visual/audible guidance would be provided. This will solve this issue and prevent a bunch of TIOs that would otherwise add complexity]]
Possible change #3. Only provide lane guidance on the four HIGHWAYS road types or at intersections with those road types (eg. primary street widens and intersects minor highway). This is just a suggestion to keep in manageable. Personally I don’t think every intersection needs to have lane control added. I think it should add value when confusion could arise.
Not sure how roundabouts can be handled. Yeah I have no idea. The only thing I can think of is using a junction box over it but not sure if that would work? Until that can be sorted I would suggest saying don’t try and use it approaching roundabouts…?
We added the use of multiple segments on ramps to effectively provide early guidance (lanes?) when coming down a ramp. Now with lane guidance this will complicate things and perhaps these forks needs to be removed and lane guidance used instead? You could possible get the best of both worlds by moving the fork all the way down to the where the ramp intersects the street so you can get the “Blah Rd to Wherever” visual and audible instruction but not lose lane guidance half way down the ramp?
I’m sure there are other issues but if we can work through these for a start then we can keep it manageable.
Cheers,
Gav…
There’s no obvious disadvantage to “over” mapping lane guidance. It doesn’t create clutter and can only serve to improve navigation. Waze even stated during the webinar they’re happy for us to map everything. So, there’s no harm mapping even small lanes, providing they meet the requirement of being a full lane at the gore point for the sake of consistency.
Adding restrictions on distance, etc overly complicates the mapping process. The only thing we’re asking right now is while you could technically map lanes on any road, please try and focus lanes on high traffic roads so when we release in Australia it helps more people. I tend to map lanes where ever I happen to be editing though.
Re the examples provided, I’m happy with LG on all those except the slip lane which doesn’t meet requirements:


Correct, I was hopeful that it would display straight prompts too. Upon reflection, I agree with Waze that this would contribute to clutter and driver distraction.
I say and post the following knowing the feature has already been announced. The “view only” and “view and hear” is currently in beta and will be released to production soon.

As above, there’s no harm, but focus on higher road types is certainly appreciated so the most value is given to Wazers when it’s released in Australia. FYI, while Waze as started to release the feature to 2% of users, Australia has the ability to control the release whenever we think we’re ready. ![]()
RA aren’t currently supported. Waze has stated they’re quite complex. They want to get the basic feature rolled out first, ensuring it meets the needs.
Lanes in junction box is also in development right now.
You might be pleased to know that alpha and beta testers contributed a lot of feedback in the early days of the feature. So much so, Waze halted and adjusted their approach to several features to meet our needs. So I’m optimistic the feature will come out later when there’s been time to think about how best to implement it.
My instinct is to say ‘yes’ get rid of all the forks, but we know it’s not going to be that easy.
The answer is we are currently experimenting in Australia. We know for heuristics to work on # and H intersections, unnecessary segments for turning lanes will cause heuristics to break. Our initial discussion on Discord and first-draft approach has been to clean up unnecessary slip lanes from intersections and add lanes. Some of us are also de-duplicating roads not meeting duplication requirements to simplify lane installation. Lanes are quite easily broken so my preference is simple roads.
The exception to this, I think, is if the way-finder provides more helpful advice than what LG would provide. An example of where way-finder info might be more helpful than LG would be Sydney airport, which gives directions to specific terminals currently.
Should we strip potentially unnecessary forks like this? Permalink. Maybe. Until we decide, I’ve been adding the lane instructions to the segment prior to the fork.
Obviously see what others think but I think it can cause a distraction when the app is popping up lane guidance for every tiny turn. Personally I would expect to see a proper “lane” with traffic in it not a tiny turning bay only big enough for one car?
I think by limiting where lane guidance is provided you reduce the map complexity that has to be maintained. Less data has to be provided to the app…
The USA Wiki is predominately focused on freeways and major highways (all the example pictures) so I don’t think it is entirely relevant to our situation where we often have “highway” road types that are only one or two lanes wide going through suburbs. Although I do note that it has “Suburban local streets typically don’t have multiple lanes, and don’t need lane guidance.”
Yes I agree with this. Regarding the link I would move the fork way down to the crossing road to get pretty much the best of both worlds…
Gav, I’m still catching up, but I would have thought lane guidance to be useful in any situation where you have to turn from a multi lane (2 or more) road?
That is unless it is only used on major roads or at intersections?
Initially I thought it would be matching the BGS (e.g. https://www.waze.com/editor/?env=row&lon=152.94561&lat=-27.57247&s=18626377727&zoom=4&segments=102019716) but then I thought it would be useful to match road markings as well (e.g. https://www.waze.com/editor/?env=row&lon=152.93755&lat=-27.51028&s=18626312191&zoom=8) . However, to have it on a single lane (e.g. https://www.waze.com/editor/?env=row&lon=152.91020&lat=-27.53035&s=18626377727&zoom=9 ) or even 2-lane road (e.g. https://www.waze.com/editor/?env=row&lon=152.91602&lat=-27.53018&s=18626377727&zoom=9) which is continuing on without signage or road markings (other than in the turning lane) just to show that there is a small turning lane to the right I didn’t think that is how it would be rolled out. Hence my thought to limit it to “lanes” of 50m or more…
Hooray Lane Guidance now working in Android Auto!