distinguish simple tls near tls controlled junctions from standalone-crossings

[ambuehll]

When I run netconvert with an osm file that has pedX and traffic signals, it generates multiple tls that I then join using tls.join. However, this does not snap them to the intersection as provided for by tls.guess-signals. Instead, it creates an extra segment in front of the actual intersection. See screenshots as well as files.

small files this time: files

BTW, on the opposite lane it also creates an additional segment from the original pedX.

SUMO-version: current build

operating system: Windows

[namdre]

This is due to your use of the legacy option --tls.uncontrolled-within which eliminates inner tls. The opposite segment is retained because it passes through the same osm node and nodes with a traffic light are not removed. If you avoid this option you get two tls on approach. If you then set option --tls.discard-simple only the tls at the stop line is kept.

Maybe it's time for you to explain why you need --tls.uncontrolled-within (most likely your are the only person in the world still using this).

[ambuehll]

Good question. I have used tls.uncontrolled-within because I got cleaner intersection tls (but I agree, that's not really a good reason to use it 👍 )

I saw that using tls.discard-simple works fine, but that in turn, ignores many other tls which I would like to keep (that's also why I started setting tls.uncontrolled-within).

So I'll just need to find out a way of keeping the other tls-simple that are relevant, since they are pedX, without a junction anywhere close by.

[namdre]

what kind of cycle behavior do you expect from those pedX? Without pedestrians they are green most of the time.

[namdre]

actually they are green most of the time regardless since we don't have #1746 yet.

[namdre]

something like --tls.discard-simple.threshold to discard tls near junctions and keep all the others.