Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [sumo-user] walking areas, od2trips, zones, and vTypes
Hello,
there is quite a difference between od2trips and duarouter in handling taz:
- od2trips uses the probabilities to pick taz edges randomly
- duarouter selects those edges from a taz which lead to the shortest route
If you want to use all edges from a taz you should use od2trips in the beginning and then continue your processing without further use of taz files.

The vehicles on top of each other are most likely due to a "vertical" edge.
As mentioned before you have some extreme elevation values in your network and this elevation causes increased edge lengths.
Hence all those vehicles fit neatly "behind" each other onto an edge which appears short in the x-y plane.

regards,
Jakob

2017-10-06 11:06 GMT+02:00 Lara CODECA <lara.codeca@xxxxxxxxxx>:
Perfect! Thank you Jakob.

"One" followup question, tho.

I used the filterDistricts.py but I still have a lot of errors using the trip from od2trips (6175 trips --> 55 routes computed, all the rest are errors). I suppose that is due to my network still to broken. Nonetheless, if I use duarouter with the flows and the filtered taz, I have routes like the following:
<vehicle id="2.89998" type="passenger2b" depart="35999.68" departLane="free" departSpeed="max" fromTaz="174561" toTaz="36990">
    <route edges="-3097#0"/>
</vehicle> 
Many of the routes have one and only edge. 
I would like to use duaIterate.py, but --districts and --additional (with the vType) cannot be used together, but if I use only additional, it does not find the taz. (solved putting the vTypes in the flows file, but I don't think this behaviour was wanted.
Nonetheless, I would like for duarouter to compute the complete route, but I don't understand how to do it. duaIterate has my same issue: during the iterations, many routes have one single edge.

The computed routes, have one peculiarity: a part from having awkward destinations (only few destination edges instead of uniformly distributed in the TAZ), all the vehicles are on top of each other (see screenshot). Is it the GUI or something is very wrong in my simulation?

Thank you again.

Lara


On 5 Oct 2017, at 23:42, Jakob Erdmann <namdre.sumo@xxxxxxxxxxxxxx> wrote:

Hi Lara,
thank you for providing the input files.
The problem was caused by pedestrian crossings with length 0 that were silently discarded by duarouter. 
The corresponding netconvert bug is fixed in revision [26352] so you need to rebuild your network.
regards,
Jakob

2017-10-04 22:08 GMT+02:00 Jakob Erdmann <namdre.sumo@googlemail.com>:
Hi Lara,
1) What is the exact error message? Which SUMO version are you using? I probably need access to your repo to investigate this.
2) You can use sumo/tools/filterDistricts.py which reads taz file, a network and a vehicle class to generate new taz definitions that only contain valid edges for that class (in revision [26330])

regards,
Jakob


2017-10-04 15:27 GMT+02:00 Lara CODECA <lara.codeca@xxxxxxxxxx>:
Dear All,

I'm building a mobility scenario from an OSM-like topology and I have various issues here and there.

1) walking areas
I'm building a scenario for pedestrian too, but if I use the options "sidewalks.guess", "crossings.guess", and "walkingareas", the resulting netfile has many of the connection tagged "walkingarea" broken (I't duarouter that points it out). Is there something that I'm missing or is it "normal" and I have to fix them by hand?

2) od2trips, zones, and vTypes
the taz are defined with a name and a list(string) of edges (all edges can be start and sink with the same probability), but in the same edge, I may have different allowed vehicles depending on the lane.
At the moment I have one complete zones definition that ignores the allowed vTypes, but when I use it with od2trips, the flows file and the trips file are both kind of unusable by duarouter or sumo. If I use taz and flows, duarouter gives me only the starting edge and it does not compute the full route, and sumo uses only one starting point for each taz. If I use the trips with duarouter, too many of them are broken because the (origins) destinations are unreachable (due to the vehicles not allowed on the edge)
Is there a simple solution for this issue? Am I using the tools in the wrong way? (probably yes)
Do I have to write different taz based on the vTypes allowed? Remember: in the same edge, I may have different allowed vehicles depending on the lane.
The last resource is to implement my own od2trips that uses only the right edges, hoping that duarouter (that is able to read my vTypes definitions) will compute the right route.

There are too many big files involved and I cannot share them on this mailing list.
In case of need, I can privately share the repo.

Thank you in advance for the help.

Regards,
Lara

---
Lara CODECA
Postdoc
Tel : 04.93.00.81.51

EURECOM
Campus SophiaTech
http://www.eurecom.fr/

_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/sumo-user






Back to the top