Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [sumo-user] Deadlocks with sublane model

Thanks Jakob!

I expected you would ask for the minimum failing example, in fact I have tried but was not able with an example built from scratch. My guess is that the behavior is due to specific intersection shape and edge geometry. So I'll work on that. Maybe I should take short edge warnings more seriously.

Of course I'll also upgrade.



On 22/04/22 14:39, Jakob Erdmann wrote:
The latest development version does indeed fix some deadlocks that could fit your description.
If you still see deadlocks with please provide a minimal failing example.

best regards,

Am Do., 21. Apr. 2022 um 15:31 Uhr schrieb joerg <schweizer.joerg@xxxxxxxxx>:
Hi all,

Using the sublane model (1m sublane width) I'm observing at some
junctions that vehicles get blocked right in front of junctions. There
seems a common pattern:

- there are 2 or 3 vehicles standing in parallel on the same lane
entering a junctions (typically 1 car+1 motorcycle or 3 motorcycles side
by side). More than 3 vehicles in parallel do not fit due to the limited
total edge width.

- vehicles seem to block each other, even though they are heading for
either  the same edge, or to different diverging edges; target edges are
free of traffic

- Something has made those vehicles stop, either a previous downstream
congestion or a traffic light signal; but after the congestion went away
or green light, those vehicles remain blocked and do not move until

- in all cases there is some divergence, either from 1 to 2 edges or
from 1 to 2 lanes.

The phenomenon generally happens in heavily congested networks. I have
see this kind of behavior already in the past, currently I am using
version 1.12

Is there any trick to resolve those conflicts or has this problem been
addressed in the latest development?

Thanks for any hint.

Best wishes,


sumo-user mailing list
To unsubscribe from this list, visit

sumo-user mailing list
To unsubscribe from this list, visit

Back to the top