Previously, wherever Waze mentioned adding support for U-Turns, our official position was that UK roads simply weren’t suitable for them. Only dual-carriageways are really wide enough to perform such a manoeuvre, and even then it may not be safe.
Now seems a good time to reevaluate this, as Waze begins testing a new U-Turn instruction. To clarify, this is a U-Turn at one end of a 2-way segment, and not on dual carriageways.
If we translate U-Turn to be a more generic “turn around” instruction, then it will include reversing around a corner and 3 point turns, or whatever is most appropriate at the time.
The ideal solution that I think that many people would like is “turn around when safe to do so”, but I fear this is beyond Waze’s ability. Also, it breaks the ETA calculations.
There are 3 main scenarios where a “turn around” instruction might be given:
Starting out, facing the wrong way
Driver missed a turning
Part of a regular route
For the first two, I think it would be reasonable for Waze to say “turn around” at a junction where an editor* has judged to be suitable. (*maybe limited to higher ranks, but there would be a huge number of junctions to consider)
I would, however, be reticent about being told to turn around as part of a planned route, just because it would save a few seconds. We’ve certainly seen duff routes in the past involving a left + u-turn instead of a right turn.
A variation on this is when a destination is only accessible from 1 direction, and requires the driver to either turn around, or “go around the houses” to get there.
I definitely think that enabled u-turns should be an exception and only set when Waze would otherwise be forced to route a user through a restriction, e.g. a private road.
I don’t think they should be enabled because a driver has missed a turn - such an event is more likely to be because the turn led into a road with a physical restriction that the driver wants to avoid. A recalculated route without a u-turn is preferable.
If we need them, it should only be in exceptional circumstances.
That would certainly be my preferred approach - its how the late 90’s era satnav in my first car behaved, so for a modern satnav not to offer it in situations where it’s the blindingly obvious preferential routing solution seems, well, a bit crap really. Admittedly, I sometimes think that old system used the “turn around” instruction as much as as a placeholder instruction whilst the system ground its way through the re-routing calculations, as a genuine instruction because it knew turning back was the better choice at the time. However, the end result was a system that did a damned good job of hiding its limitations whilst giving instructions that generally felt sensible. How many human navigators would, the instant their driver took a wrong turn, start frantically working out a completely new route, and how many would just go “uhh, we wanted the right turn back there, pull into that layby and spin around”…
And ETA calculations be damned, if attempting to maintain a reliable ETA is cause for the routing engine to generate a “continuation” route which is longer (time or distance-wise) than the “turn-about” route, then IMO it’s a poor design decision. Being provided with reliable ETAs is one of the selling points of Waze, sure. However, there are times when it really is OK to inject a bit of uncertainty if, in doing so, it opens up the ability to provide a better result to the user.
Most of the time I don’t mind that Waze isn’t necessarily giving me the shortest/fastest route that it’s aware of there and then, if it believes that the route it’s actually offered me is more likely to get me to my destination within a few minutes of the stated ETA. However, if I slip up and take a wrong turn somewhere, what I don’t want it to do is then immediately come up with what might be a completely different route simply because it’s still trying its damndest to give me a reliable ETA, and especially not if that route is one which adds noticeably to the journey time/distance.
For case 1, I’d be happier if Waze always assumed the user was pointing in whichever direction allowed for the optimal route to be selected. If we’re not going to get an explicit “turn around where possible” instruction in the app, we could at least be given an implied one in this situation where chances are quite high that you’d be able to do so. In the background, the app could still also have calculated the optimal route based on the assumed starting direction of travel, and switch over to suggesting that route if the user ends up driving that way for long enough whilst trying to find somewhere to turn around such that it’d now be better for them to simply continue.
Case 2 I’ve touched on above, and as with 1, if we’re not going to get a turn around instruction, the same logic could still apply here - leave the original route displayed until turning back is no longer the shortest/fastest route.
That I agree with 100%. The “u-turn”/“turn around” instruction should only be made available to the routing engine as the first instruction of a route (as in cases 1&2 above), it shouldn’t be used as a means to further optimise the remainder of the route.
The idea of a verbal instruction to turn round at a dead-end is very sensible, but doesn’t need a ma edit. Even this scenario has problems though; if I’m starting a route facing the wrong way, I will want to turn round as soon as possible rather than drive all the way to the end before doing so.
I also know of several examples where there is no space to turn at the end of a road so you must turn round before then - or be faced with reversing out again.
There are other examples where you must turn round, e.g. road ahead is private segment or restricted to buses only, but again I think Waze should be able to give a ‘turn around’ instruction without us needing to do anything in the maps.
I don’t see why this should be outside the capabilities of Waze developers - I think that the current issues around "turn around’ instructions need to be fixed in the Waze routing engine and not in the map data.
As Twister-UK says, when starting out, offer me the best route in any direction.
This is what you get on the Live Map website. I have this problem every time I route from home as there are Waze often excludes the best route as it incorrectly believes I’m facing a particular direction - in fact I’m parked on the drive and can turn either way.
The natural reaction to taking a wrong turn is to turn around and go the right way.
This is not a particularly difficult algorithm to implement based on the additional time needed to take a new route in the wrong direction.
Waze know when you went ‘off-route’ and what the remaining drive time was. Ideally Waze could instruct you to ‘turn around where possible’ until such point as the new route is quicker.
The other problem here is that the decision to ‘turn around’ needs to be that of the drivers rather than the editor. It may be easier to turn round in a lay-by or entrance to premises rather than at the next junction. I am usually happy to do so in my car, other’s don’t like performing such maneuvers, or may be towing a trailer - or perhaps the traffic is too heavy and for whatever reason therefore prefer to continue on the new route.
The only valid example I can thing of is when U-turns are allowed on dual-carriageways in order to access a side street that cannot be directly accessed across the central reservation. These can be handled by adding a turn segment or using a bow-tie junction which gives us the ability to enable the ‘U-turn’.
Unless and until Waze implement something sensible, there is a work-around available to map editors only. Simply add a stub “private road” for your drive - 5m will be long enough - and then ensure that your “home” destination is nearer to that stub than the public road. Keeping it short ensures that it won’t be noticeable to others on the app map.
Today, unknowingly until I got home, I was taken on a 20 mile detour because waze didn’t take in to account I could turn either direction from the point I was stopped at. Of course I could have checked the route, but it was a 450 mile marathon round trip in unfamiliar terrain so I was more than happy to let Waze take the strain of deciding where to go.
So, if not U-turns, Waze changing their algorithm to allow routing in either direction from a standing stop on a two way road.
Waze (and for that matter most other SatNav’s) can have trouble defining what direction you are pointing in when you start a journey. If you are pointing in the correct direction OR if Waze thinks you are pointing in the correct direction it will give a route with out any diversions or detours. If you aren’t (or Waze doesn’t think you are) you will get some strange routing at the beginning of your route. If you are unfamiliar with where you are or don’t have a map to check which general direction you should be heading in then that type of detour will be a regular occurrence.
It may be a point worth bringing up with Waze HQ to set route calculation from your location and start giving the route you are best taking AND assuming you are heading in the correct direction to start with?
I get this when starting from home all the time. I am directed to a mini roundabout to turn around and go in the right direction!
I agree with both Twister-UK and PealRinger for 1, 2 & 3.
As Waze has routing penalties for various situations could this not be the way of implementing ‘turn around where possible’ for situation 2 - if I missed my turn I would happily take a new route if it added a minute or so to my journey but if it added 10 minutes I would be trying to turn around.
I don’t know how the routing penalties work (does anyone outside Waze HQ?) but could it be that each situation has a different value of routing penalty, something like.
Are you finding them in the map, or being used for routing in the client? Currently, U-turns aren’t used for routing, but other countries have started to test them with proper support in the client. I started this discussion to see if there are any valid use cases here in the UK.
When the report was raised, the u-turn at the junction node between the public and private segments had been disabled, and so Waze had no alternative but to provide a route that continued through the private section and onto Beacon Way. In the same situation now, although the client gives slightly odd-looking instructions when the u-turn is required, the route it provides does at least avoid using the private section of Shepherds Way, which is the desired result from the user perspective.
This type of scenario is one that wouldn’t exist if private segments were handled properly when calculating routes, rather than simply slapping a routing penalty onto them and hoping that another route can be found with a lower overall score… Ideally the private segment would be treated as being non-existent, so that the routing server would then see the remaining part of Shepherds Way as a virtual cul-de-sac and generate the same “proceed to highlighted route” instructions you get with a real cul-de-sac.
No, directing the user up to the u-turn-enabled node just to turn back isn’t the most appropriate solution in an ideal world, however given the Waze system limitations we’re working within at the moment, in these types of scenario it’s certainly preferable to having Waze generate a route through what ought to be an inviolable restriction.
IMO Waze should never direct a user to perform a maneouvre immediately unless it’s certain that the maneouvre can be performed at that exact location. So giving an instruction to turn at an explicitly marked junction is OK, but giving an instruction to do a u-turn there and then isn’t. Deciding where to do a u-turn is something best left to the user, and any instructions/directions/guidance/etc given by Waze should reflect that - my preference would therefore be for something like “turn around where possible”.
I am noticing that quiet a few AM’s are starting to use U-Turns (my original question wasn’t directed at Twister).
I simply wanted clarification on the current stance.
I can see exactly why Twister has done this and it’s a fix that works - however perhaps we should consider for the time being allowing only Lv 4 / 5 / 6 to ability to use the U-turn where appropriate - I am not even sure that this is possible!!!
I agree with Twister totally on this point - it’s a bit like the old adage - if I tell you to jump off a cliff would you do it… the sad answer is some would, therefore the phrase “where possible” indicates that some poor drivers might have to revert to A:- thinking B:- using the windscreen (oh the trouble that will cause a few).
Not only is this a situation where we could use the u-turn function, I would say this is where we absolutely should. Not that there are many examples of this, I should think.
