[sumo-user] Output differences between TraCI & libsumo for mesosim

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

[sumo-user] Output differences between TraCI & libsumo for mesosim

marcelreppi
Hello,

I have a custom TraCI scenario and for performance improvement I wanted to
test how much faster libsumo and mesosim are in my case.
I am only interested in the routes vehicles take and how much distance they
have driven.
I would expect that my output data is the same when using TraCI or libsumo
with/without mesosim.
I wanted to make sure that my output is not affected by some unexpected
behavior so I collected some data and realized that there are quite some
differences.

I have some polygons which I call zones and I track the vehicle movement in
these zones.
The driven distances are sometimes very different between the different
execution settings (TraCI/libsumo with/without mesosim).

I have attached a picture with my collected data.
I tested two scenarios:
    - "n" which is just regular SUMO with only periodic rerouting
    - "rr" which is TraCI with periodic rerouting AND custom rerouting

Is my expectation wrong? Am I missing something?

Any help is appreciated!

Thanks!

Best regards
Marcel

<http://sumo-user-mailing-list.90755.n8.nabble.com/file/t349/sumo-comparison.png>




--
Sent from: http://sumo-user-mailing-list.90755.n8.nabble.com/
_______________________________________________
sumo-user mailing list
[hidden email]
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
Reply | Threaded
Open this post in threaded view
|

Re: [sumo-user] Output differences between TraCI & libsumo for mesosim

behrisch
Administrator
Hi Marcel,
there should be no difference between traci and libsumo but since traci
for meso is a quite recent addition there might still be some bugs. Can
you share your scenario?

Best regards,
Michael

Am 24.06.20 um 18:10 schrieb marcelreppi:

> Hello,
>
> I have a custom TraCI scenario and for performance improvement I wanted to
> test how much faster libsumo and mesosim are in my case.
> I am only interested in the routes vehicles take and how much distance they
> have driven.
> I would expect that my output data is the same when using TraCI or libsumo
> with/without mesosim.
> I wanted to make sure that my output is not affected by some unexpected
> behavior so I collected some data and realized that there are quite some
> differences.
>
> I have some polygons which I call zones and I track the vehicle movement in
> these zones.
> The driven distances are sometimes very different between the different
> execution settings (TraCI/libsumo with/without mesosim).
>
> I have attached a picture with my collected data.
> I tested two scenarios:
>     - "n" which is just regular SUMO with only periodic rerouting
>     - "rr" which is TraCI with periodic rerouting AND custom rerouting
>
> Is my expectation wrong? Am I missing something?
>
> Any help is appreciated!
>
> Thanks!
>
> Best regards
> Marcel
>
> <http://sumo-user-mailing-list.90755.n8.nabble.com/file/t349/sumo-comparison.png>
>
>
>
>
> --
> Sent from: http://sumo-user-mailing-list.90755.n8.nabble.com/
> _______________________________________________
> sumo-user mailing list
> [hidden email]
> To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
>


_______________________________________________
sumo-user mailing list
[hidden email]
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user

signature.asc (201 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [sumo-user] Output differences between TraCI & libsumo for mesosim

Jakob Erdmann
If you use the same network, there will be differences in the recorded distances (and thus travel times) due to the way junctions are treated differently. See https://sumo.dlr.de/docs/Simulation/Distances.html#no-internal-links_meso

Am Sa., 27. Juni 2020 um 22:40 Uhr schrieb Michael Behrisch <[hidden email]>:
Hi Marcel,
there should be no difference between traci and libsumo but since traci
for meso is a quite recent addition there might still be some bugs. Can
you share your scenario?

Best regards,
Michael

Am 24.06.20 um 18:10 schrieb marcelreppi:
> Hello,
>
> I have a custom TraCI scenario and for performance improvement I wanted to
> test how much faster libsumo and mesosim are in my case.
> I am only interested in the routes vehicles take and how much distance they
> have driven.
> I would expect that my output data is the same when using TraCI or libsumo
> with/without mesosim.
> I wanted to make sure that my output is not affected by some unexpected
> behavior so I collected some data and realized that there are quite some
> differences.
>
> I have some polygons which I call zones and I track the vehicle movement in
> these zones.
> The driven distances are sometimes very different between the different
> execution settings (TraCI/libsumo with/without mesosim).
>
> I have attached a picture with my collected data.
> I tested two scenarios:
>     - "n" which is just regular SUMO with only periodic rerouting
>     - "rr" which is TraCI with periodic rerouting AND custom rerouting
>
> Is my expectation wrong? Am I missing something?
>
> Any help is appreciated!
>
> Thanks!
>
> Best regards
> Marcel
>
> <http://sumo-user-mailing-list.90755.n8.nabble.com/file/t349/sumo-comparison.png>
>
>
>
>
> --
> Sent from: http://sumo-user-mailing-list.90755.n8.nabble.com/
> _______________________________________________
> sumo-user mailing list
> [hidden email]
> To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
>


_______________________________________________
sumo-user mailing list
[hidden email]
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user

_______________________________________________
sumo-user mailing list
[hidden email]
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
Reply | Threaded
Open this post in threaded view
|

Re: [sumo-user] Output differences between TraCI & libsumo for mesosim

marcelreppi
Thanks for your replies!

I have made another test which also showed the difference in values for meso
and micro.
The results are attached.

@Jakob: I am already using "--no-internal-links true" in my regular TraCI
scenario so I expected them to be equal.

@Michael: Here is the network and demand data for my scenario:
https://u.pcloud.link/publink/show?code=XZn0UIkZr33jVAq6TbYQ2GyyQrAVjRUgI4Dk

My TraCI code is quite complex so I did not include it.
If you still need it I can try to reduce it to the core and send it to you.

Thanks for your efforts!

<http://sumo-user-mailing-list.90755.n8.nabble.com/file/t349/sumo-comparison.png>





--
Sent from: http://sumo-user-mailing-list.90755.n8.nabble.com/
_______________________________________________
sumo-user mailing list
[hidden email]
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
Reply | Threaded
Open this post in threaded view
|

Re: [sumo-user] Output differences between TraCI & libsumo for mesosim

Jakob Erdmann
In your first post you mentioned that your scenario involves rerouting. Depending on the used options for the meso junction model the traffic states can differ significantly between meso and micro. This could lead to different routes and thus different distances. One way to investigate this would be to generate edgeData output for micro/meso, compare those with tools/output/edgeDataDiff.py and then look at the differences in sumo-gui. The next step would be to observe the actual simulations at those spots that differ most in edgeData metrics such as 'speed'.

Am Mi., 1. Juli 2020 um 14:49 Uhr schrieb marcelreppi <[hidden email]>:
Thanks for your replies!

I have made another test which also showed the difference in values for meso
and micro.
The results are attached.

@Jakob: I am already using "--no-internal-links true" in my regular TraCI
scenario so I expected them to be equal.

@Michael: Here is the network and demand data for my scenario:
https://u.pcloud.link/publink/show?code=XZn0UIkZr33jVAq6TbYQ2GyyQrAVjRUgI4Dk

My TraCI code is quite complex so I did not include it.
If you still need it I can try to reduce it to the core and send it to you.

Thanks for your efforts!

<http://sumo-user-mailing-list.90755.n8.nabble.com/file/t349/sumo-comparison.png>





--
Sent from: http://sumo-user-mailing-list.90755.n8.nabble.com/
_______________________________________________
sumo-user mailing list
[hidden email]
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user

_______________________________________________
sumo-user mailing list
[hidden email]
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user