Does cuOpt natively support Multiple Trips / Vehicle Reloading at the Depot?

[mariomartinezgovoy]

I am working on a Vehicle Routing Problem (VRP) where vehicles have limited capacity. Once a vehicle finishes delivering its current load, it needs to return to the depot, reload, and go back out to continue fulfilling the remaining orders. In optimization, this is typically known as the Vehicle Routing Problem with Multiple Trips (VRPMT).

Does NVIDIA cuOpt currently support this behavior intrinsically or natively?

Any examples or guidance on how to approach this scenario would be highly appreciated.

Thanks!

[rg20]

@mariomartinezgovoy cuOpt does not natively support this feature.

This can however be modeled as a pickup-delivery problem. Can you try that approach?

[mariomartinezgovoy]

Thanks for the suggestion! I get the general concept, but could you share some tips or guidance from your experience on how to properly set this up?

[np96]

@mariomartinezgovoy hi there.

PDPTW modeling will not guarantee that vehicle load will be drained each run.
To model repeated depot reloads in one solve, create pickup-delivery pairs where every pickup order uses the depot location id and each delivery order uses its customer location. Pickup/delivery pairs are order indices, not location ids. Use positive demand on depot pickups and matching negative demand on deliveries; vehicle capacity controls how many deliveries can be loaded per depot visit. Use set_skip_first_trips() and set_drop_return_trips() when the initial departure and final return legs should not be charged, while intermediate depot pickup tasks still represent reloads.

vrpmt-example.py
Running the attached script should produce such output:

Status: 0
Vehicle-0 starts at: 0.0, completes at: 88.0, travel time: 88.0,  Route : 
0(P)->2(P)->1(D)->3(D)->4(P)->6(P)->5(D)->7(D) 

This results in a travel time of 88.0 to deliver all routes


   route  arrival_stamp  truck_id  location      type
0      0            0.0         0         0    Pickup
1      2            2.0         0         0    Pickup
2      1           14.0         0         1  Delivery
3      3           31.0         0         2  Delivery
4      4           51.0         0         0    Pickup
5      6           53.0         0         0    Pickup
6      5           75.0         0         3  Delivery
7      7           88.0         0         4  Delivery

[mariomartinezgovoy]

Thank you very much Nikolai!