Interesting behaviour on a pre-planned route that uses the same road in both directions

[paul69]

This is for information only and perhaps for others to relate this behaviour. I have sent the log files to the devs for investigation.
I've attached the route below so others can review it and determine whether this was user error. On this occasion, I do not believe it was, but I'm happy to be proven wrong.
https://www.myrouteapp.com/route/open/14195502

Setup
iPhone 13 Pro running the latest NN beta version
Carpuride 702 (no issues observed with the device itself)
Offline navigation enabled
Automatic recalculation enabled (I don't normally use this, but it was enabled specifically for this test)
Route as tracks enabled
All maps downloaded and up to date

The route was all good until I was travelling from waypoint 4 to waypoint 5.
Along this section, I passed a junction on my right. This junction was part of the route later in the day, where I would make a left turn towards waypoint 12 when travelling in the opposite direction.
As soon as I passed this junction while heading towards waypoint 5, NN recalculated the route and began insisting that I turn around and navigate to waypoint 12.
I could not use Skip Waypoint, as doing so would have sent me to waypoint 13, which was not correct.
As I knew the planned route, I ignored the instructions. NN continued trying to send me to waypoint 12 as I travelled through Hathersage.
When I eventually stopped to investigate, I noticed that waypoints 6 to 11 were greyed out, as though they had already been visited, despite the fact that I had not yet reached them.
To recover, I long-pressed waypoint 6 and selected it as my next point. NN then resumed navigation and I continued riding.

After passing waypoint 7, NN then attempted to direct me to waypoint 11, which was now displayed as the next waypoint.
Again, I ignored the instructions and continued along the planned route, passing waypoints 7 to 11 in sequence.
Once I had reached waypoint 11, NN resumed normal operation and correctly navigated the remainder of the route (12 → 13 → 14, etc.).

The same behaviour happened again on another section of the route where the outbound and return legs used the same road.
After passing waypoint 17, I passed a junction on my right that I would use later when returning along the route.
As soon as I passed this junction, NN repeatedly instructed me to turn around and navigate directly to waypoint 22.
These instructions continued all the way into Macclesfield.
I eventually stopped, long-pressed waypoint 18, and selected it. NN then immediately resumed the correct navigation instructions and followed the original planned route.

[paul69]

I am going to ride this route again, with the iPhone set up the same way and I will also use my RT3, the settings will be the same for both, but RT3 with no internet.
Will report back with my findings

[Nick Carthew]

Do you have Automatically skip waypoints on, it would help to trace the issue if you do have this selected to turn it off for your next test. Don't adjust the route, keep it exactly the same as the route you shared.

[Hubert Thoring]

I think it's due to the automatic recalculation, which I usually turn off myself.

I would recommend after the test as Nick has described also to adjust the WP and closer to the turn.

So SP 5 closer to the junction maybe 100m, and for this SP 12 further away from the junction about 300m ... so it is perhaps better clear to the navigation system that SP 5 is closer than SP 12 is ...

The same applies to SP 17, and certainly to SP 18 compared to SP 22.

When turning, it is always better to set a SP in 100m after the turn.

Have fun testing. VG

[paul69]

hi @Nick-Carthew

it was turned on, not sure why, TBF I usually have this turned off,
but surely this should not affect the route in reality, why would the algorithm not pass waypoints in order, does seem strange behaviour.

therefore, possibly when planning routes one will have to ensure that a later way/shaping point is not to close to a current one, another thing to think about jeez!!

will see what today brings

[paul69]

hi @Hubert-Thoring said:

So SP 5 closer to the junction maybe 100m, and for this SP 12 further away from the junction about 300m ... so it is perhaps better clear to the navigation system that SP 5 is closer than SP 12 is ..

SP 12 is 400m from the junction, as I've stated above, it would seem we need to ensure later SPs are further away than upcoming current ones if using same roads on a return journey. Does seem to be another thing we need to think about when planning routes.
If this is the case then this thread will definitely give some tips to other users.

The same applies to SP 17, and certainly to SP 18 compared to SP 22.

Sp 17 is 900m away from junction, I do not need SP18 to be any closer as the route only really has one way to go.

When turning, it is always better to set a SP in 100m after the turn.

not sure why this is the case, or do you mean at least 100m, anyway, I am aware of the need to not place them close to a turn and I use this method in all my planning.

[Nick Carthew]

@paul69 You are correct, it should not affect the route in the way that it has. The developers are looking into improving the algorithm for automatic skipping but it really isn't that easy. There are a lot of things to take into consideration when writing the new algorithm.

[Con Hennekens]

@paul69 said:

As soon as I passed this junction while heading towards waypoint 5, NN recalculated the route and began insisting that I turn around and navigate to waypoint 12.

This "normal" behavior, in case there is a recalculation. The algorithm checks max 10 route points ahead for which one is closest, and obviously nr 12 is candidate for that. It's true that when there are short loops like this, this is not always the intended way, but it's a choice between bads... BUT there are two mechanisms that should have prevented this:

• why is there a recalculation in the first place, when you never took a wrong turn?
• if this was the first recalculation you should have been guided to #5 anyway

In my view, if you did not take a wrong way, there should have been no recalculation and the loop would have been completed normally. But in case you actually did something to trigger multiple recalculations, this behavior is exactly like the algorithm is programmed

Thanks for sharing, because these experiences is what's needed to refine the algorithm.

[paul69]

hi @Con-Hennekens said:

The algorithm checks max 10 route points ahead for which one is closest, and obviously nr 12 is candidate for that.

that is probably the only explanation for this, #12 is indeed the closest at that point of the junction, it will certainly be something to watch out for in the future when planning routes and the use of shaping points.

@Con-Hennekens said:

why is there a recalculation in the first place, when you never took a wrong turn?
if this was the first recalculation you should have been guided to #5 anyway

I didn't take a wrong turn and once passing that junction I did not receive any instructions toward #5, even after passing over #5, I only had continuous instructions to turn around.
After resetting to #6 and passing over that one, all I then got were instructions to #11, very odd. but I have an update will do it in another post.

cheers @con

[paul69]

Hi @Nick-Carthew said:

Do you have Automatically skip waypoints on, it would help to trace the issue if you do have this selected to turn it off for your next test. Don't adjust the route, keep it exactly the same as the route you shared.

well I rode the route the again this afternoon, well up to #12, the route itself was not changed

iPhone behaviour,
all settings the same save for turning off "automatically skip waypoints"

All went well passed 1 to 4 with no issues headed toward the junction and yes you guessed it, no recalculations and it just continued as the planned route onward to 5, 6, 7 etc.
So I then stopped the route, rode back to 500m before #4 and started again from nearest waypoint which was #4 obviously, this time I reverted to turning on "automatically skip waypoints" as this was the setting the previous day.
I then rode over #4 toward #5 passed the junction in question and, there were no instructions to turn around an head for #12. aaaarrrggghhhh!!!!

RT3 running the same route with same settings, this device does not have any mobile data/5G

all instructions matched the iPhone until I reached the junction between #4 and #5, this device then carried out the turn around instructions to return to #12 as I received yesterday on my iPhone.

this has bamboozled me now, I may have to carry out some more testing using this small route.