A shipment leaving a factory in Groningen, Netherlands, actually takes several days and hops across multiple vehicles before it reaches a consumer in Versailles, via Utrecht, Antwerp, and Paris. 📦 In logistics optimization, "first mile" (producer to hub) and "last mile" (hub to consumer) have been studied for decades. But the segment connecting them, "middle mile," relaying goods between distribution centers at regional or continental scale, has stayed largely unstudied despite its huge operational cost. The reason is simple: logistics companies guard their network topology and demand volumes as trade secrets, leaving researchers without realistic data to work with. 🧭 Google Research's "MilleMiglia: A Realistic Instance Generator for Middle-Mile Logistics" ( fills that gap. Instead of treating middle-mile as just another VRP variant, it reframes it as a multi-commodity flow problem on a space-time graph, folding three non-negotiable hard constraints, fixed schedules, hub throughput limits, and arrival-triggered synchronization, into a single unified file format. 🌍 Using gravity-model hub placement and realistic demand distributions, it generates everything from small academic toy problems to continent-scale industrial instances and large ML training datasets, all while protecting privacy. The goal: become the CVRPLIB of middle-mile research, a standard benchmark the whole community can build on. #LogisticsOptimization# #OperationsResearch#

How do goods travel 450 miles overnight? It’s thanks to careful logistics optimization, particularly in the middle-mile segment, but optimizing these networks is difficult without public data. MilleMiglia provides a standardized benchmark, using spatial clustering and gravity models to simulate realistic scenarios for middle-mile delivery problems. Learn more: