C.H. Robinson beat its old routing tech on 10 benchmarks with one framework

C.H. Robinson (CHR), a global freight provider, faces 'rich' vehicle routing problems daily: each customer has distinct objectives and constraints.

CHR previously relied on third-party routing technologies, each tuned to a specific problem type, which meant long setup times for new customers.

Researchers proposed a set-partitioning framework with several route-generation algorithms; it outperformed the existing technologies at CHR on 10 benchmark instances.

The framework, wrapped in a simple user interface, was integrated into CHR's transportation planning and execution platform, reducing setup times and improving planner productivity and customer outcomes.

One framework, many route generators

One framework reused algorithms across many problem types.

Winning 10 benchmark instances gave the evidence needed to replace incumbents.

A simple UI lowered the barrier to daily use.

Faster setup turned new customer onboarding from weeks to days.

Instead of buying a different routing engine per customer, build one framework that plugs in generators per constraint set; the platform improves instead of the per-case workaround.

Since implementation, the module has outperformed the previously used third-party technologies, significantly reduced setup times and improved users' productivity as well as customer outcomes; the work was published in INFORMS Journal on Applied Analytics (2022) and as an arXiv preprint.