Enhancement Request for Vehicle Routing Problem / Last Mile Delivery Solutions

[Krishnendu_MB]

Enhancement Request for Vehicle Routing Problem / Last Mile Delivery Solutions

While using the Vehicle Routing Problem (VRP) and Last Mile Delivery solutions in ArcGIS, the current workflow primarily depends on known routes and randomly allocates orders to depots based on these routes. However, in real-world scenarios such as online retail or postal services, orders are already pre-assigned to specific depots based on predefined serviceability zones.

At present, ArcGIS Network Analyst tools do not offer functionality to incorporate this dependency — i.e., the mapping of orders to depots — into the routing process.

We would like to propose the following enhancements to better reflect real-life operational requirements:

1. Support for Order-to-Depot Mapping: Enable route optimization based on the existing mapping of orders to depots or hubs, rather than assigning orders dynamically.

2. Constraint-Based Route Generation: Allow users to specify the number of delivery personnel (i.e., the number of routes) available per depot. The solution should then generate optimal routes within each depot’s service area, respecting the predefined order-to-depot assignments.

These capabilities would significantly improve the relevance and applicability of ArcGIS solutions for industries with fixed delivery territories and structured logistics models.

[MingyiHe18]

Thank you for your suggestions! I’d like to clarify that both features you mentioned are currently supported in our solver:

1. Assigning orders to depots: While there is no direct preassignment between orders and depots, you can achieve this functionality by either preassigning orders to specific routes or by using the Route Specialties and Order Specialties tables. This ensures that orders are served by routes that originate from the desired depot. You may also split your vrp or last mile delivery problem to multiple problems based on order - depots to get faster solving.

2. Limiting the number of routes from a depot: This can be configured using the Add Fleet Routing Routes tool. You can specify both the number of routes and their corresponding start/end depots. For example, you may add 1 route to Depot1 and 2 routes to Depot2 by using the tool twice .

Please feel free to let me know if you have any further questions!

[Krishnendu_MB]

Hi @MingyiHe18 , 

Thank you for the detailed clarification and for pointing out the existing capabilities.

1. I had explored the use of Route Specialties and Order Specialties earlier. While this does help in guiding the solver, the primary requirement in our case is to have routes as an output rather than a predefined input. For retail companies, the objective is often to let the solver determine the optimal routes based on fixed order-to-depot assignments — being able to predefine the depot but not the route structure itself. Having a more native way to incorporate order-to-depot mapping without needing to preassign routes would be highly beneficial and align more closely with operational workflows.

2. As for limiting routes from depots, while the Add Fleet Routing Routes tool provides the needed control, the process becomes repetitive and manual — especially in production environments where daily route generation is required based on varying staff availability. A more streamlined or automated mechanism to define route counts per depot dynamically would greatly improve scalability and usability in real-world deployments.

I hope these points help convey the intended enhancement more clearly. Appreciate your support and open engagement on this!

[MingyiHe18]

@Krishnendu_MB 

Thank you for your thoughtful response and additional clarification. To further elaborate:

1. Routes as Input and Output: Routes serve as both inputs (constraints) and outputs (solutions). Initially, route constraints must be provided to initiate the solver, after which the solver generates the optimal selection and structure of routes.

In both vehicle routing problem and last mile delivery solvers, "Depots" specifically refer to locations where routes originate and conclude. Regarding your point about letting the solver determine optimal routes based on fixed order-to-depot assignments, here are two typical scenarios:

• Many-to-Many Scenario: If multiple depots and orders exist, and routes are expected to connect depots with orders, you can add depots into the Orders sublayer and use Order Pairs table within the Vehicle Routing Problem (VRP) solver to establish "pick-up and drop-off" relationships between orders. Additionally, VRP allows routes to start directly from an order location without explicitly associating them to a depot.

• One-to-Many (or Several-to-Many) Scenario: When orders significantly outnumber depots, the Last Mile Delivery solver is typically the most suitable option. As depots explicitly represent the start and end points for routes, orders assigned to a depot will naturally translate into a depot-to-order relationship. The order specialty and route specialty methods in my last reply could solve the problem.

2. Automation and Scalability: For automated and scalable workflows, the ArcGIS Python API provides robust support. For more detailed information, please refer to the following resources: https://mediaspace.esri.com/media/t/1_u4can3v9/238781913, https://www.esri.com/content/dam/esrisites/en-us/about/events/media/UC-2019/technical-workshops/tw-6...