[sumo-user] unexpected problem and how to find suitable indicators to evaluate routing algorithm

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

[sumo-user] unexpected problem and how to find suitable indicators to evaluate routing algorithm

Jane Cheung
Dear all,
As a small test (attachment), I find the shortest routes by using duaroute.
While I was using the shortest as the route-file in .sumocfg and active the re-routing command and set the probability to '1'.

The statistic results are confused me a little.
The statistic values after rerouting are larger(e.g., duration, waiting time). 
And the inserted vehicles are less. 
As described, the re-routing command finds faster routes for departure vehicles. 
If there is none fastest route, vehicles should keep the original routes. The duration should not be larger than without the re-routing device.

How to explain this appearance, Or how to avoid this unexpected problem? 
Best regards,

Jane








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

case9.zip (38K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [sumo-user] unexpected problem and how to find suitable indicators to evaluate routing algorithm

Jakob Erdmann
Your scenario highlights one of the deficits of the rerouting device: It cannot look into the future.
The algorithms for estimating travel speed / travel time look into the past and therefore have no reliable information about roads which are unused. This makes the detour via empty roads look promising even though they are not helpful.
There are some ideas on how to improve this:

Note, that in many urban scenarios, most roads see some traffic most of the time so that the "no-information" condition is encountered less often.

Am Di., 8. Sept. 2020 um 17:50 Uhr schrieb Jane Cheung <[hidden email]>:
Dear all,
As a small test (attachment), I find the shortest routes by using duaroute.
While I was using the shortest as the route-file in .sumocfg and active the re-routing command and set the probability to '1'.

The statistic results are confused me a little.
The statistic values after rerouting are larger(e.g., duration, waiting time). 
And the inserted vehicles are less. 
As described, the re-routing command finds faster routes for departure vehicles. 
If there is none fastest route, vehicles should keep the original routes. The duration should not be larger than without the re-routing device.

How to explain this appearance, Or how to avoid this unexpected problem? 
Best regards,

Jane







_______________________________________________
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] unexpected problem and how to find suitable indicators to evaluate routing algorithm

Jakob Erdmann
Also, you may find it helpful to color edges in sumo-gui by "routing device assumed speed" to understand what is happening.

Am Mi., 9. Sept. 2020 um 07:46 Uhr schrieb Jakob Erdmann <[hidden email]>:
Your scenario highlights one of the deficits of the rerouting device: It cannot look into the future.
The algorithms for estimating travel speed / travel time look into the past and therefore have no reliable information about roads which are unused. This makes the detour via empty roads look promising even though they are not helpful.
There are some ideas on how to improve this:

Note, that in many urban scenarios, most roads see some traffic most of the time so that the "no-information" condition is encountered less often.

Am Di., 8. Sept. 2020 um 17:50 Uhr schrieb Jane Cheung <[hidden email]>:
Dear all,
As a small test (attachment), I find the shortest routes by using duaroute.
While I was using the shortest as the route-file in .sumocfg and active the re-routing command and set the probability to '1'.

The statistic results are confused me a little.
The statistic values after rerouting are larger(e.g., duration, waiting time). 
And the inserted vehicles are less. 
As described, the re-routing command finds faster routes for departure vehicles. 
If there is none fastest route, vehicles should keep the original routes. The duration should not be larger than without the re-routing device.

How to explain this appearance, Or how to avoid this unexpected problem? 
Best regards,

Jane







_______________________________________________
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] unexpected problem and how to find suitable indicators to evaluate routing algorithm

Jane Cheung
Jakob, 
Thanks for your reply! 

Appreciated

Jane

On Wed, Sep 9, 2020 at 1:47 PM Jakob Erdmann <[hidden email]> wrote:
Also, you may find it helpful to color edges in sumo-gui by "routing device assumed speed" to understand what is happening.

Am Mi., 9. Sept. 2020 um 07:46 Uhr schrieb Jakob Erdmann <[hidden email]>:
Your scenario highlights one of the deficits of the rerouting device: It cannot look into the future.
The algorithms for estimating travel speed / travel time look into the past and therefore have no reliable information about roads which are unused. This makes the detour via empty roads look promising even though they are not helpful.
There are some ideas on how to improve this:

Note, that in many urban scenarios, most roads see some traffic most of the time so that the "no-information" condition is encountered less often.

Am Di., 8. Sept. 2020 um 17:50 Uhr schrieb Jane Cheung <[hidden email]>:
Dear all,
As a small test (attachment), I find the shortest routes by using duaroute.
While I was using the shortest as the route-file in .sumocfg and active the re-routing command and set the probability to '1'.

The statistic results are confused me a little.
The statistic values after rerouting are larger(e.g., duration, waiting time). 
And the inserted vehicles are less. 
As described, the re-routing command finds faster routes for departure vehicles. 
If there is none fastest route, vehicles should keep the original routes. The duration should not be larger than without the re-routing device.

How to explain this appearance, Or how to avoid this unexpected problem? 
Best regards,

Jane







_______________________________________________
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