A parcel has connections to catch
Google’s September 18, 2026 MilleMiglia overview explains a middle-mile planning problem: freight moves between distribution centers and may switch vehicles. A late arrival can miss an outgoing connection. The model must account for vehicle schedules, sorting capacity and the order in which transfers become possible. [1]
- Reach the hub: is the parcel on time?
- Catch the vehicle: is space available?
- Complete the transfer: can the hub cope?
Generate problems before comparing answers
MilleMiglia generates synthetic logistics instances for comparing optimization methods and training machine-learning models. Its public README describes generated hub graphs, paths through those hubs and assigned travel durations. Those choices define the problems a tested method gets to see. [2]
A shorter path can lose
Hypothetical example: A parcel reaches Hub B at 10:05, after the 10:00 truck has left. A longer route through Hub C could arrive sooner at the destination if its connection is still available. These invented times illustrate why total distance alone does not settle a transfer problem.
Go a little deeper
Optional reading · about 1 more minute
Realistic is a claim to examine
The developers say the generator models spatial patterns, demand and structured schedules using statistical distributions informed by public and privately disclosed information. That describes their design approach; it does not establish that every generated instance matches a particular carrier. The linked paper was inaccessible during our review, so we do not report paper-level validation or operational savings. [1]
What an evaluator should ask
Our interpretation: Before comparing scores, ask whether the test includes the constraints that could reverse a decision: a full vehicle, a missed departure or limited sorting capacity. A method that wins on an easier representation may not answer the same operational question. The benchmark itself deserves scrutiny alongside the solver.
Original sources
Attributed synthesis, not original reporting. Examples labeled hypothetical or illustrative are explanatory. Reviewing a source does not independently validate its findings.
- Google Research: MilleMiglia ↗
September 18, 2026 developer overview read September 30. Linked HAL paper blocked; no bypass or paper-validation claim.
- MilleMiglia repository ↗
Public README Goal and Generation Details read September 30, 2026. No code execution or independent solver comparison.