Periodic Reporting for period 1 - ACUMEN (Ai-aided deCision tool for seamless mUltiModal nEtwork and traffic managemeNt)
Période du rapport: 2023-06-01 au 2024-11-30
Modern urban mobility faces unprecedented challenges as cities struggle to manage increasingly complex multimodal transport systems while meeting ambitious sustainability, safety and efficiency goals. The emergence of new mobility services like on-demand transport, Mobility-as-a-Service (MaaS), e-scooters, and e-cargo bikes, is fundamentally disrupting how citizens and businesses move, making traditional traffic management approaches increasingly inadequate. Despite advances in Machine Learning, AI, and Connected Mobility technologies, their full potential remains untapped due to fragmented implementation, lack of coordination between modes and operators, and absence of frameworks for data exchange and interoperable decision-making.
ACUMEN is transforming this landscape by developing a generic, privacy-preserving, data-driven digital paradigm for advanced network management. The project's vision is to support the transition to seamless, sustainable, connected and automated mobility through three key innovations: (1) designing a secure, decentralised data framework enabling real-time information sharing between mobility providers; (2) leveraging explainable AI and hybrid intelligence for unprecedented accuracy in monitoring and forecasting; and (3) developing novel decision-making solutions that foster cooperation between mobility providers across all urban scales.
These innovations are being integrated into a digital twin environment, offering stakeholders a unified view of transport systems to better handle both daily operations and long-term evolution. The project is demonstrating impact through four complementary pilots in Athens, Helsinki, Amsterdam and Luxembourg, each testing different aspects of the system under real conditions. Expected impacts include reduction in waiting times for multimodal trips, dea crease in network congestion, and improvement in public transport level of service.
ACUMEN takes an interdisciplinary approach, combining technical expertise in transport engineering, computer science and AI with social sciences perspectives on user behaviour, policy frameworks and governance models. This integration ensures the developed solutions address not just technical challenges but also social, economic and institutional barriers to adoption.
ACUMEN is transforming this landscape by developing a generic, privacy-preserving, data-driven digital paradigm for advanced network management. The project's vision is to support the transition to seamless, sustainable, connected and automated mobility through three key innovations: (1) designing a secure, decentralised data framework enabling real-time information sharing between mobility providers; (2) leveraging explainable AI and hybrid intelligence for unprecedented accuracy in monitoring and forecasting; and (3) developing novel decision-making solutions that foster cooperation between mobility providers across all urban scales.
These innovations are being integrated into a digital twin environment, offering stakeholders a unified view of transport systems to better handle both daily operations and long-term evolution. The project is demonstrating impact through four complementary pilots in Athens, Helsinki, Amsterdam and Luxembourg, each testing different aspects of the system under real conditions. Expected impacts include reduction in waiting times for multimodal trips, dea crease in network congestion, and improvement in public transport level of service.
ACUMEN takes an interdisciplinary approach, combining technical expertise in transport engineering, computer science and AI with social sciences perspectives on user behaviour, policy frameworks and governance models. This integration ensures the developed solutions address not just technical challenges but also social, economic and institutional barriers to adoption.
ACUMEN has established a secure and privacy-preserving data framework that leverages federated learning capabilities. The consortium has successfully developed AI-based data fusion models that combine multiple data sources, including low-cost Bluetooth data and drone-collected trajectories, to estimate passenger numbers in public transport. A real-time incident detection system using unsupervised learning techniques has also been implemented, enhancing the ability to identify and respond to network disruptions. Additionally, a Dynamic Graph Convolutional Recurrent Neural Network has been developed for multimodal forecasting, which will be integrated into the anomaly detection system.
For forecasting and simulation, ACUMEN has advanced the development of demand forecasting models that exploit heterogeneous data sources while incorporating traffic flow theory to ensure trustworthy predictions. A significant achievement has been the creation of an auto-calibration component that enables fast and accurate calibration of simulation models, essential for real-world applications.
The traffic management aspects have progressed through the development of a systematic framework for generating stress test scenarios, which helps identify critical factors that could lead to network failures. The project has implemented various management strategies, including trajectory pricing and perimeter control, while also developing an AI-based proxy modeling framework utilizing Reinforcement Learning techniques for efficient decision-making.
A cornerstone technological achievement has been the development of the ACUMEN Digital Twin platform, built on a microservices architecture. The platform incorporates core services for data management, API gateway functions, and scenario management. Significant progress has been made in integrating various simulation tools, enabling comprehensive testing and validation of management strategies. The platform has also successfully integrated the previously described data fusion models, forecasting components, and traffic management tools, creating a cohesive environment for multimodal network management.
ACUMEN has successfully initiated implementation across four pilot sites. In Athens, extensive data collection has been completed using drones across nine locations, integrated with other traffic data sources. The Helsinki pilot has established multiple data collection systems and begun testing incentive-based traffic management approaches. Amsterdam has focused on developing detailed simulation networks and implementing multi-modal traffic signal control strategies. In Luxembourg, an on-demand shuttle service has been launched, accompanied by the development of an on-demand service application.
These technical achievements have generated substantial scientific output, with multiple conference papers and journal articles either published or under review. The integration of various technological components, from data collection to decision support systems, demonstrates ACUMEN's progress toward its goal of enabling seamless multimodal network management. The successful implementation across different pilot sites validates the practical applicability of the developed solutions while providing valuable insights for further refinement.
For forecasting and simulation, ACUMEN has advanced the development of demand forecasting models that exploit heterogeneous data sources while incorporating traffic flow theory to ensure trustworthy predictions. A significant achievement has been the creation of an auto-calibration component that enables fast and accurate calibration of simulation models, essential for real-world applications.
The traffic management aspects have progressed through the development of a systematic framework for generating stress test scenarios, which helps identify critical factors that could lead to network failures. The project has implemented various management strategies, including trajectory pricing and perimeter control, while also developing an AI-based proxy modeling framework utilizing Reinforcement Learning techniques for efficient decision-making.
A cornerstone technological achievement has been the development of the ACUMEN Digital Twin platform, built on a microservices architecture. The platform incorporates core services for data management, API gateway functions, and scenario management. Significant progress has been made in integrating various simulation tools, enabling comprehensive testing and validation of management strategies. The platform has also successfully integrated the previously described data fusion models, forecasting components, and traffic management tools, creating a cohesive environment for multimodal network management.
ACUMEN has successfully initiated implementation across four pilot sites. In Athens, extensive data collection has been completed using drones across nine locations, integrated with other traffic data sources. The Helsinki pilot has established multiple data collection systems and begun testing incentive-based traffic management approaches. Amsterdam has focused on developing detailed simulation networks and implementing multi-modal traffic signal control strategies. In Luxembourg, an on-demand shuttle service has been launched, accompanied by the development of an on-demand service application.
These technical achievements have generated substantial scientific output, with multiple conference papers and journal articles either published or under review. The integration of various technological components, from data collection to decision support systems, demonstrates ACUMEN's progress toward its goal of enabling seamless multimodal network management. The successful implementation across different pilot sites validates the practical applicability of the developed solutions while providing valuable insights for further refinement.
ACUMEN has shown several promising results in its first period. The project has successfully developed and implemented core technical components including the data framework, AI-based fusion models, and the Digital Twin platform, which are being tested across four pilot sites. These developments have generated substantial scientific output through conference papers and journal articles.
To ensure further uptake and success, the project has identified several key needs through its work with stakeholders and pilot implementations. The collected data from pilots, particularly in Helsinki and Athens, demonstrates the importance of continued validation in real urban environments. The project's Reference Group, representing both public and private sectors, has been instrumental in identifying practical implementation requirements.
The governance workshops organised as part of the project, with one already completed in Luxembourg and others planned in Helsinki, Athens, and Amsterdam, are helping to define the necessary regulatory and standardization frameworks. These workshops are particularly focused on AI integration into traffic management systems, addressing both technical and governance aspects.
The exploitation plan outlines methodologies for identifying exploitable results and managing intellectual property rights, which will be crucial for future commercialization. The plan emphasizes sustainability and long-term impact, ensuring innovations remain relevant beyond the project lifecycle. The project has also established mechanisms for knowledge dissemination, indicating growing interest from potential stakeholders and users.
To ensure further uptake and success, the project has identified several key needs through its work with stakeholders and pilot implementations. The collected data from pilots, particularly in Helsinki and Athens, demonstrates the importance of continued validation in real urban environments. The project's Reference Group, representing both public and private sectors, has been instrumental in identifying practical implementation requirements.
The governance workshops organised as part of the project, with one already completed in Luxembourg and others planned in Helsinki, Athens, and Amsterdam, are helping to define the necessary regulatory and standardization frameworks. These workshops are particularly focused on AI integration into traffic management systems, addressing both technical and governance aspects.
The exploitation plan outlines methodologies for identifying exploitable results and managing intellectual property rights, which will be crucial for future commercialization. The plan emphasizes sustainability and long-term impact, ensuring innovations remain relevant beyond the project lifecycle. The project has also established mechanisms for knowledge dissemination, indicating growing interest from potential stakeholders and users.