Re: [sumo-user] Vehicles not used in drtOnline output

[Jakob Erdmann <namdre.sumo@xxxxxxxxx>]

Hello,

The script is known to consume quite a bit of compute time and has built-in timeouts to abort the solution search.

This might be the reason for not making use of all vehicles.

On the other hand it might be a bug in the algorithm.

Unfortunately, the original developer of the algorithm is not actively developing it anymore so I cannot give you a time frame for having this issue analyzed or fixed.

I recommend the following

- try running the tool with different timeout option values to see whether that affects vehicle utilization

- try to create a minimal use case for unused vehicles which you are able to share (i.e. so that we can put it in our test suite)

- consider taking a deeper look into the algorithm implementation to check whether you can find the source of the problem yourself

In the meantime I'll forward your questions to the developer of drtOnline to obtain more insights.

regards,

Jakob

Am Mi., 12. Jan. 2022 um 15:59 Uhr schrieb François Poinsignon <fr.poinsignon@xxxxxxxxx>:

Hi,

I am a student working on a master thesis project the integration of autonomous vehicle fleets to public transport.

I am trying to use the drtOnline tool (see https://sumo.dlr.de/docs/Tools/Drt.html) to simulate an autonomous fleet service in a small city but I am not sure if I am using it correctly. Since I found relatively little information about it online, I am asking here for your help.

I followed the instructions given in the documentation and noticed a strange behavior: there seems to be a threshold on the number of vehicles added to the system beyond which additional vehicles are not used anymore (0 customers, 0 occupied time and distance). I imagined that adding more vehicles would lead to a decrease of the average customer waiting time and maybe to the average customer trip duration as well.

It also happens that some customers are not served while some vehicles are not being used at all. This makes no sense to me as not serving a customer has a very high cost by default.

For example, if I have 100 customers and 20 cars as an input, the output will be that 5 customers are not served and that the 20 cars are actually used. But if I input 25 cars instead, again 5 customers will not be served, and 5 cars will not be used at all.

I also have an additional question: Are there easy ways to decrease the simulation time? In my case, running the tool with about 1000 customers and maybe 200 vehicles already takes several hours.

Thank you for reading this, any help would be greatly appreciated.

François

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