ACUMEN

Modern urban mobility systems are becoming increasingly complex due to the growth of multimodal transport, shared mobility, logistics, and real-time digital platforms. Although large volumes of mobility data are available, current transport systems remain fragmented, relying on siloed data sources and limited coordination across modes, operators, and jurisdictions. This results in inefficiencies in traffic management, weaker disruption prediction, and reduced capacity to respond to events such as accidents, demand surges, or infrastructure failures. Increasing requirements for sustainability, accessibility, and resilience further intensify pressure on urban mobility systems, which must operate under stricter environmental and operational constraints.

Previous research in transport modelling, forecasting, and digital infrastructure has addressed parts of these challenges, particularly through data analytics, simulation, and early digital twin approaches. However, these solutions often lack system-level integration and cannot fully capture multimodal demand–supply interactions in real time. In addition, many AI-based methods suffer from limited explainability, weak generalisation to new conditions, and poor alignment with transport theory and operational decision needs. These gaps highlight the need for a unified, scalable, and interpretable framework for next-generation mobility systems.

The ACUMEN project proposes a step change in mobility management through a federated Digital Twin framework integrating multimodal data sources, AI/ML models, and hybrid intelligence mechanisms. It combines real-time processing, predictive modelling, and simulation tools to enable door-to-door mobility optimisation for passengers and freight.

The project introduces a Hybrid Intelligence approach where humans and AI systems jointly support decision-making across strategic, tactical, and operational levels. This includes explainable machine learning enhanced by sensitivity analysis for improved interpretability and robustness under uncertainty. Methods such as reinforcement learning, federated learning, and multimodal data fusion are integrated into a unified framework to capture transport demand and supply dynamics.

LIST leads the development of the ACUMEN Digital Twin, a microservices-based platform combining data management, APIs, simulation, visualisation, and decision support. It integrates AI tools from other work packages and connects simulation engines such as SUMO, MATSim, MnMs, and Aimsun Next via standard APIs. Overall, WP5 provides a scalable, interoperable environment for real-time mobility management and simulation-driven decision-making across pilot sites.

ACUMEN will deliver interoperable Digital Twin components, multimodal forecasting and simulation tools, and data-driven decision support systems for real-time mobility management. These outputs will enable traffic anomaly detection, improved prediction of demand and supply dynamics, and better coordination across transport modes and stakeholders. The platform will also provide advanced visualisation capabilities to support the interpretation of complex mobility scenarios.

From an industrial perspective, the project is expected to improve the efficiency, resilience, and sustainability of urban transport systems through better infrastructure utilisation, reduced congestion, and enhanced multimodal coordination. Transport authorities, mobility providers, and logistics operators will be able to optimise fleet management, improve service reliability, and respond more effectively to disruptions.

ACUMEN could support adaptive traffic control, dynamic public transport scheduling, real-time disruption management, and integrated urban logistics optimisation. It also lays the foundation for future smart city infrastructures, where digital twins interact continuously with physical systems to enable predictive and prescriptive mobility management at urban and regional scales.