Re: [sumo-user] '--no-internal-links' suppress the insertion of new vehi

[Jakob Erdmann <namdre.sumo@xxxxxxxxx>]

When simulating without internal links then the end of an incoming lane is directly connected with the start of an outgoing lane and there a vehicle can be half on either lane as you conjectured correctly.

However, even in this case, cross traffic can still flow across the intersection (because the intervening intersection space is not modeled).

Because cross traffic can always flow, there is no code to prevent vehicles from getting stuck halfway on the intersection.

In contrast, when simulating with internal links, there is an extra behavior module that prevents driving onto the intersection when the downstream link has insufficient space.

Note, that gridlock can always happen if a traffic jam loops back on itself. (A simple case would be a roundabout that is completely filled and has come to a standstill.

When no vehicle can reach the desired exit you have a deadlock. The same applies to bigger meshes in the road network.

regards,

Jakob

Am Do., 19. März 2020 um 18:30 Uhr schrieb Hao Zhou <zhouhaoseu@xxxxxxxxxx>:

Hi Jakob, thanka for your answer. Later on I found the solution is to set 'depart speed' 'depart pos' and 'depart lane'.

Now I am able to generate high density with '--no-internal-links'. However, it is still puzzling to me why gridlock still exsit, and more specifically, I noticed some vehicle can be half on origin lane, and maybe half on the next lane. As the figure shows, the vehicle is already on southbound lane D6D5_0, but its origin eastbound lane E6D6 is still blocked. I conjecture that the vehicle is still half located on its origin lane. Do you have any ideas to fix that?

I know a manual method, which is to check downstream availability for every lead vehicle using traci in every simulation step. If the availability is not enough, manully pause the vehicle by setting its speed to zero. I think the method is not efficient. Hope you have a better solution. 

Thank you and take care,

Hao

On March 19, 2020 at 12:00 AM, Jakob Erdmann <namdre.sumo@xxxxxxxxx> wrote:

Hello,

the sumo option --no-internal-links switches to a model where the lanes within intersections are no longer used. This reduces the total road space of the simulation and can thereby affect insertion (vehicle insertion can be delayed/fail due to lack of road space).

If you set the netconvert option --no-internal-links when building the network, then the length of "normal" (non-intersection) edges is artificially increased to compensate for the missing road space.

Note, that some types of blockage (jam on intersections) is impossible when using --no-internal-links.

regards,

Jakob

Am Mi., 18. März 2020 um 15:46 Uhr schrieb Hao Zhou <zhouhaoseu@xxxxxxxxxx>:

Hi Sumo community,

Hope everyone is safe and healthy.

I encountered an interesting problem where vehicles cannot be inserted to the network if I set '--no-internal-links' to be true. Does this option suppress the insertion of vehicles?

My goal is to keep generating new vehicles using traci.vehicle.add to reach a very congested network. It can be done with internal links, but as expected we will see vehicles entering the intersection and lead to gridlock. When setting '--no-internal-links', the vehicle accumulation is significantly slower and as the figure shows, there seems a limit on the number of vehicles.   

Does anyone have similar experience? Thank you in advance!

Best,

Hao Zhou,

Georgia Tech

_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user

_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user

_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user