Optimization
Plan freight across firms, not one fleet at a time.
We think freight should be planned across firms and modes, because fleets that each plan well can still waste fuel together. Trucks return empty and loads travel half-full when no one looks at the whole flow. Eurostat reports that 21.8% of the kilometers driven by road freight vehicles in the EU in 2025 were driven empty. The IPCC found that road vehicles produced 70% of direct transport emissions in 2019. extendfuture would model freight flows and fuel use, so fleets and governments can test changes such as shared depots or a shift to rail before they make them.
How can fleets and governments cut fuel use in freight?
Fuel use falls when the same goods travel fewer and fuller kilometers on the right mode. We would model where freight starts and ends, which roads, railways and ports it uses, and how full vehicles run. Open-source routing and optimization tools then test changes such as shared depots, return loads, delivery time windows or a shift to rail. AI agents, software that sets up runs and drafts summaries, compare fuel, cost and emissions for each. The people who run the fleets and set the policies decide what changes.
Plan the whole flow, not one fleet
Each firm can plan its own trucks well and the total can still waste fuel, because no one plans the flow between firms. Governments carry part of the bill. The UN's 2026 SDG report counts more than USD 921 billion of fossil fuel subsidies worldwide in 2024 and says such subsidies divert limited public finance away from sustainable development. Less fuel burned means lower costs for firms and, where fuel is subsidized, a lower subsidy bill for governments.
Start with public statistics, add fleet data by choice
A first model can map freight flows from published statistics. It would cover freight flows between regions, the road, rail and port network, vehicle types and their fuel use, and the depots where loads are combined.
- Road freight statistics from national statistics offices and Eurostat, including empty running
- Road and rail networks from OpenStreetMap
- Fuel sales by region and vehicle registrations from official statistics
- Port and rail freight volumes published by operators and transport ministries
- Telematics, meaning GPS and engine data, from fleets that choose to share it
Measure the saving against today's pattern
A saving counts only if it is measured against what trucks do today. We would first rebuild today's freight pattern, meaning the tonnes moved between regions, the mode used and the fuel burned. The model is checked against published totals for empty running, fuel sales and tonne-kilometers, the measure of one tonne carried one kilometer. Routing and loading are then optimized with tools such as OR-Tools and VROOM. Every saving is reported with its assumptions, such as how many firms would share a depot.
Test a plan on real days before anyone commits
Transport ministries, city logistics teams and fleet operators would each get the view they need, and fleets would try new plans on real days first.
- Fuel, cost and emissions for each scenario, such as a shared consolidation center or a shift of freight to rail
- Maps of the corridors with the most empty running
- Route and loading plans that fleets can test on a sample of real days
- The effect of a delivery time window or a low-emission zone on fuel use and delivery cost
- A plain-language comparison of scenarios, checked by an analyst
Fleet data is for routes and loads, not for ranking drivers
Fleet data is commercially sensitive. Each fleet's data stays separate, and published results show only totals that cannot be traced to one firm. Driver data is used for routes and loads, not to rank or discipline drivers. Scenarios are checked for who carries new costs, such as small carriers facing a new rule. Transport officials and fleet managers decide what changes, and every run is logged.
Build on open tools that every firm can check
A plan shared between competing firms needs tools that each of them can inspect. Each tool below is open source or open data.
- OR-Tools, Google's open-source optimization suite for vehicle routing, scheduling and linear and integer programming
- VROOM, an open-source engine for vehicle routing problems with capacities, time windows and pickup and delivery
- OSRM, a routing engine that runs on OpenStreetMap data and computes travel times between many points
- OpenStreetMap, for roads, rail lines, ports and depots
- SUMO, whose tools include emission calculation, for fuel and emission estimates on a simulated road network
- statsforecast, for forecasting freight volumes and fuel demand
Is this for governments or for fleet operators?
Both. A fleet operator can start with its own routes and loads, and a ministry or city with regional freight flows. The two views can meet in a shared scenario, such as a consolidation center at the edge of a city.
How much fuel could we save?
We do not quote a number in advance. The first model compares today's routes and empty running with what the optimizer finds on your own data, and reports the gap with its assumptions. That gap is the honest estimate.
Does it cover rail, shipping and air freight?
Rail and ports are part of the network and the mode choice. Shipping and air freight enter as flows through ports and airports, not as detailed vessel or flight models.
Can it help plan electric trucks?
Yes, as a scenario. The model can test which routes suit battery trucks given range and charging points, and what charging adds to grid load, which links to the energy model.
Are drivers tracked?
Only as fleets already track vehicles, and only for routes and loads. The model does not score individual drivers, and published results show totals only.
Talk to the people who build.
One call. An honest read on what AI can do for this, and the number it has to beat.