search

eclipse-sumo / sumo

Eclipse SUMO is an open source, highly portable, microscopic and continuous traffic simulation package designed to handle large networks. It allows for intermodal simulation including pedestrians and comes with a large set of tools for scenario creation.
https://eclipse.dev/sumo
Eclipse Public License 2.0
2.53k stars 1.43k forks source link

Deadends for private vehicles due to lane permissions #13586

Open gladysmcgan opened 1 year ago

gladysmcgan commented 1 year ago

I'm picking up from the previous discussion #13087, sorry to bring it up again!

I'm still facing a similar issue where connections cannot be accessed by private vehicles because the target lane is a dedicated lane. This causes private vehicles to have deadends in the network.

In the example here, the first lane of the incoming north edge connects to a bus lane.

Current connections: <img width="360" alt="Screenshot 2023-07-06 at 6 07 03 PM" src="https://github.com/eclipse/sumo/assets/59922884/3d1b2aeb-a0f4-433f-b450-2ba9ec52ad0f">

Desired connections:

Screenshot 2023-07-06 at 6 06 45 PM

A potential solution is perhaps that, as a rule, lanes preferentially do not connect to subsequent lanes with fewer permissions, but rather the lane with the same or more permissions. Buses from the preceding edge can switch lanes afterwards to go onto the bus lane.

Also, it seems like the bike lanes are 'taking up' the connections that would make more sense for the regular lanes. Maybe the bike lanes can have a different rules for connection building, with regard to merging/zipping into a regular lane, since bikes do not have the same issue as regular vehicles with merging into one lane.

The same (but reversed) problem in a simpler intersection:

Screenshot 2023-07-06 at 6 15 00 PM

Any help or ideas for a workaround is appreciated! Many thanks :)

Example intersections in here: google drive folder

SUMO version 1.17

m-kro commented 1 year ago

Essentially the merging/zipping rules are the problem. Your potential solution may fit for your particular intersection but not for others... We have to think a bit more how to solve this kind of problems without introducing more elsewhere.

m-kro commented 1 year ago

@behrisch Do you have an idea how to make the OSM import rules evolve in this case?