During RP2, GREEN-LOG achieved major advancements beyond the state of the art in urban logistics, data sharing and multimodal last-mile optimisation. The project moved from design to deployment of integrated platforms tested across five Living Labs, demonstrating innovative tools and governance models in real conditions. A key result is the second version of the City Logistics Data Space, a federated FIWARE-based environment enabling real-time, semantically interoperable data exchange across operators and cities. Enhanced access control and automated data-ingestion pipelines were implemented, ensuring operational readiness. GREEN-LOG delivered significant progress in forecasting and optimisation. The AI-driven demand prediction and Nowcasting & Forecasting Engine integrate environmental, operational and behavioural datasets and use event-driven retraining to improve accuracy. The project routing and optimisation service handling ~600-parcel datasets, enabling TSP-based clustering and supporting collaborative logistics among competing LSPs. Early demonstrations showed profitability gains in Athens and more efficient mobile-depot deployment reducing travel times in Oxfordshire, demonstrating clear benefits over static routing tools. The first integrated versions of the LaaS Marketplace and DCT were also available in RP2. The Marketplace, piloted with 58 users in Mechelen and supporting 497 deliveries, integrates dynamic pricing and nudging to promote sustainable choices. The DCT integrates all WP3 services and provides secure authentication, real-time dashboards, policy constraints and incentive modules, combining city-level governance with operator-level intelligence. The BYOD app, redesigned through user feedback and integrated with the Data Space and routing engine, enables low-cost, vendor-neutral device incorporation. The Augmented Intelligence Modelling Platform and simulation service extend the state of the art through multi-city mesoscopic simulation (Oxford, Athens, Barcelona) and support complex what-if analyses tied to real data. Early results show measurable CO2, cost and time reductions and strong user acceptance across LLs, confirming the feasibility of shared, multimodal and digitally governed last-mile ecosystems.
