pumrum wrote:

tedtrost wrote:An alternative idea would be that each segment could have a checkbox called "Divided", and the system would know that drivers on the wrong side could not turn into houses on the other side if that box is checked.

I like that idea even better. I know Waze had a similar feature in the past, but I wasn't around so I don't know how if differed from your idea, if at all.

You're correct, this was a past feature which was supposed to do exactly what your OP wants. Problem was that Waze didn't implement house numbers for years and because we often had junctions into locations at parking lot entrances, just because the house number was on one side of the divided street, didn't mean you couldn't get to it from the left.

I'm not sure how Google nav pulls this off, but your idea sounds reasonable.

pumrum wrote:Any downsides to tedtrost's idea of adding a property to each two-way segment that indicates "divided"? Having it be per segment would allow you to break the segment in to two segments if there was a break in the median, or if the median ended.

Yes. We'd have to add "extra" junctions wherever a median which restricts left turns (right in some parts of the world) and that would add a lot more "objects" on the map, adding to complexity of the map, and potentially, depending on density, decreasing app performance because it's got more objects. (not a lot, but it will.)

A normally straight stretch of single-segment road, might had two or three medians in it. That means 5 or 7 segments instead of 1.

I think the best we can do for now, is to "hope" that the business is far enough off the street that we can put in even a small parking lot segment or two, which are closer to the lat/lon of the location than the street. With turn restrictions set properly at the street, we can basically force the same thing, causing Waze to route through the parking lot segment to get there, which means it should follow the restrictions and arrive at the location from the right side of the street.

This doesn't work for all locations, such as those with no off-street parking and for houses, for which we do not map driveways. In those cases, that situation is enough reason to split into divided roadways. Again, not the ideal, but we have to balance map complexity with usefulness, and sometimes we have no choice but to make it a bit more complex. And, of course, we have to work within the confines of the current system functionality.

Let me be more explicit as a precursor to a Bugzilla writeup.

Add a check box for segments titled "Divided Road" (similar to toll road).

With a two-way road if the median was down the entire block, you have one segment with this "divided road" check box. Waze routing would default to houses on the "right" side of the street in "right-hand drive" countries require routing that has them headed from the end of the block that permits driving on that side of the road. If coming the other way it would then require the u-turn at the end of the block.

If there is a split in the middle of the block to allow a turn around just add one junction and enable u-turns there. Both segments on either side of that junction would still have "divided road" checked. Houses that can be reached by turning at that junction would be routed through that u-turn. Otherwise they go to the end of the block to the u-turn (or other routing if needed).

If there are multiple splits in the middle of the block, do you really need to tell the driver where to turn around? It would be like having a center turn lane on any other road where the driver has to make a decision where in the center turn lane to pull in and turn around. If there are other cars to avoid then the driver makes those decisions. I think for that case you tell editors to not mark divided road and let drivers find one of the many turn slots between the median.

Personally in small residential blocks with medians, I was never worried much about this issue since you can typically get to the other side of the street visually. You can see where you go to the end of the medial and turn around. But in larger cities with many "no u-turns" like San Francisco, this can be a nightmare, which is why we have been splitting some of the streets up there. With this options we could go back to a two-way street.

Alan, once we get general consensus on this solution, I propose we write this up for Bugzilla to get it in there for consideration in case the Devs are thinking of something different that may not work as well (but will be too deep to change from it). I am happy to enter it if you don't beat me to it (once we have it fairly fleshed out.)

Is there more we are forgetting around this?

pumrum wrote:Excellent write-up, thank you! The assumption here is that Waze implements U-Turn handling. Otherwise everything is clear and concise.

Yes. It generally takes months before big changes/additions like this make it into the system, so we will tie in the requirement for u-turns to be operational when this goes live.

russblau wrote:It sounds an awful lot like this proposal.

Indeed. It sounds like we are proposing the same element and implementation. We can link that topic in the Bugzilla for additional reference.

Just found there is a current Bugzilla and I added this thread to its comments. Vote if you have access.

That is a good idea, but would require a large amount of editing.

An alternative idea would be that each segment could have a checkbox called "Divided", and the system would know that drivers on the wrong side could not turn into houses on the other side if that box is checked.

kentsmith9 wrote: If there is a split in the middle of the block to allow a turn around just add one junction and enable u-turns there.

If legally permitted. Otherwise, no u-turn, and Waze will route to the best way to turn around (could be the next break with a legal u-turn, a jug-handle, or go around the block.