3.2.14. Auto skip feedback
-
Just did a very quick and crude test on my way to work this morning to test out the newly re-introduced Auto Skip feature which I really liked previously. Although it had its flaws in the previous/current Nav app, most of the time it worked a charm.
Unfortunately some flaws in the previous algorithm are still present in the 3.2.14 version. I created this ridiculous route to test it, imho the algorithm should be able to handle this very fine...
Red is overlayed tracklog.If for whatever reason, any next waypoint is closer than the waypoint to skip, the algorithm should auto skip all waypoints up to the currently closest waypoint. Currently this causes the following issue:
Note how I'm already near waypoint 11, while the algorithm is still in skipping mode, trying to skip waypoint 6. This screenshot was taken while driving near 11, but the issue already occurred when driving near 8. Apparently the app cannot skip multiple waypoints at once, or it only compares the current location with the next waypoint, relative to the one being currently skipped. I.e.: when skipping wp 6, only wp 7 is 'looked at' to see whether it is closer, while any waypoint beyond wp 6 should be taken into consideration.
This is especially needed if for some reason big stretches of road are blocked, or some small loops for which many waypoints were required, cannot be driven.
Generally I like the skip feature already! The button is big enough to revert the auto skip with gloves on. Much better than the current app! I do however liked the 'countdown' indicator that shows the rider how long he has time to revert or cancel the autoskip. Can we get that back?
One issue in the current app that I did not encounter yet in the beta version, is that it seems to identify a instruction as ignored, when the rider did not change its bearing, while receiving the instruction eventhough the rider didn't even have a chance to adhere to the instruction. What that means is that for instance a waypoint is missed, but after that waypoint a very big stretch of straight road is driven on. Even though the app provides instructions to 'turn around', one simply cannot since the road does not lend itself for that. From my memory, the current app already identifies the instruction as 'ignored' when the next instruction to do the same is given again, without having the rider given the chance to actually driving PAST the piece of the road, that he or she would be able to turn around at, i.e. a roundabout. Imho the app should only identify a instruction as ignored, if the rider actually has ignored/driven past the road circumstance that facilitates the instruction to be fulfilled.
I do realize while typing this feedback though, that if my comment above regarding the next waypoints to be taken into account, will be implemented, this actual problem might not exist anymore.Manks bu'j te bange.
-
@StefanHummelink Excellent feedback! Thanks a lot for this elaborate test.
What I can say is that this scenario is exactly something we also tested during the development.
We are not a great fan of letting the algorithm route to any nearest point. Why? For the test route you created, it makes sense. However, for a very different route (most likely a roundtour) there is a chance that the closest waypoint might be the one on the way back - resulting in skipping a lot of waypoints you don't want.
Yes, the above scenario can be countered with:
- You should use some viapoints (some people simply don't)
- The algorithm may only skip a maximum of x points (but there is no correct assumption for x. 10? 20?)
So in your situation / test where the route is actually a great test but not a realistic route, the algorithm does exactly what it should. And if this occurs, the only real (and safe) option is to skip a bunch of waypoints yourself.
Hope this gives a good insight in my opinion about
If for whatever reason, any next waypoint is closer than the waypoint to skip, the algorithm should auto skip all waypoints up to the currently closest waypoint
We simply won't do that as there can be too many drawbacks to that. Leaning to the safe side is better in this case. But indeed, for your specific test case, there wouldn't be these drawbacks
Your other points
- The countdown is something we wish to add too
- Regarding your other more complex comment, we would need to investigate that. Not really sure what it is about.
-
@Corjan-Meijerink said in 3.2.14. Auto skip feedback:
We are not a great fan of letting the algorithm route to any nearest point. Why? For the test route you created, it makes sense. However, for a very different route (most likely a roundtour) there is a chance that the closest waypoint might be the one on the way back - resulting in skipping a lot of waypoints you don't want.
not sure if this is possible, but what about checking direction too? If I'm moving, e.g. north and the closest waypoint has a "south direction" you can ignore it.
The algorithm may only skip a maximum of x points (but there is no correct assumption for x. 10? 20?)
what about skipping a maximum of x kilometers?
I haven't used myroute yet to navigate, but having to click on my phone while riding my motorbike (or stopping to do so) isn't something I'd really like to do...
The current app I use (which I'd like to stop using because of other bugs) currently seems to handle autoskip quite nicely and it seems it doesn't get confused about skipping waypoints on the way back. If you want I can write you privately the name in case you could be interested in doing some benchmarks and / or trying to reverse engineer its algorithm. -
@cvlmtg Thanks for the input!
We know exactly how the previous app implemented the behavior so that's settled
Luckily we were involved in the development But yes, good suggestion indeed! We could indeed limit the skip to x kilometers so you don't suddenly skip 200km of route.
I like the discussion already going on here. Let's have a drive with the app and get some more actual tests. Feedback like this is indeed exactly what we are looking for!
We do tend to like a general assumption that (test) routes are realistic
-
@Corjan-Meijerink said in 3.2.14. Auto skip feedback:
@cvlmtg Thanks for the input!
We know exactly how the previous app implemented the behavior so that's settled
Luckily we were involved in the development I think there has been some misunderstanding, because I never talked about the previous myroute app
-
@cvlmtg Aaah, I interpreted the "current app I use" as "the current app" (where app would be Navigation)
-
@StefanHummelink said in 3.2.14. Auto skip feedback:
I.e.: when skipping wp 6, only wp 7 is 'looked at' to see whether it is closer, while any waypoint beyond wp 6 should be taken into consideration.
I am not so sure this is preferred behavior. There is no saying that when I skip e.g. WP6 and WP11 is the closest at the moment, that I don't want to follow 7, 8, 9 and 10. Especially in routes that have crossings / loops in them, that would lead to a very early beer at the destination...
I don't think we should skip multiple WP's at all. I DO think we should skip an extra WP if we are moving closer to WP n+2 so to say. So skipping WP 6 routes you to WP7, and only if WP 8 becomes closer than WP 7 we would skip 8 and be directed to 9.
I am not affiliated with MyRoute-App. The help I am offering is entirely voluntary.
I am just an enthusiastic user, and hope you will be one too!Most motorcycle problems are caused by the nut that connects the handlebar to the saddle.
-
@Con-Hennekens Great input! Personally I agree with this logic too
-
@Corjan-Meijerink said in 3.2.14. Auto skip feedback:
But yes, good suggestion indeed! We could indeed limit the skip to x kilometers so you don't suddenly skip 200km of route.
Great nuance, I can see that a waypoint up to 500m / 1km ahead would be much better than my suggestion to skip all. Lol, thinking back to it, I should not have stated it that bold, but hey, at least we got a decent discussion starting here.
Regarding the situation where this issue might arise which I actually encountered:
Due to roadwork, I was unable to drive through 2, 3 and 4 and before 2 was actually skipped, I already passed wp 5 (not in the picture). This situation kept going and going... Had to stop at the side of the road. In this situation the app should be clever enough to know that 3 and 4 should have been skipped immediately as soon as I've passed 5.Somehow the app should look to not only the upcoming waypoints but also to the fact whether or not, the rider is still on the intended route, albeit further along then what should have originally been the case.
Regarding my second rather complex comment, I can provide some graphical explanation but perhaps I should just do some more testing to check whether this issue actually still exists.
-
@Con-Hennekens said in 3.2.14. Auto skip feedback:
I don't think we should skip multiple WP's at all. I DO think we should skip an extra WP if we are moving closer to WP n+2 so to say. So skipping WP 6 routes you to WP7, and only if WP 8 becomes closer than WP 7 we would skip 8 and be directed to 9.
Awesome improvement of my suggestion... I acknowledge that I was a bit crude.
Simply wanted to get the feedback out there too fast... "Haastige spoed is zelden goed", seems to be appropriate again...Manks bu'j te bange.
-
@StefanHummelink, not at all, We all need suggestions to come up with even better suggestions. Hopefully someone comes up with an even better suggestion than mine, because I can benefit from that one too!
-
We’ll get there! Glad we all like the first version of the algorithm
It will be improved where possible based on all suggestions!
-
Referenced by
Corjan Meijerink
