VDI/VDE Innovation + Technik GmbH as part of the stakeholder engagement activities of the FAME project organised, on 18 November, a Live Talk on how data and AI are transforming the CCAM ecosystem.
Moderated by Oihana Otaegui Madurga (Director of Transport & Security Division, Vicomtech), the session bridged the gap between application-driven solutions and the more transversal, generic and research-oriented AI advancements.
Among the panellists, Margriet van Schijndel, Program Director Responsible Mobility, TU Eindhoven, CCAM Partnership, and member of the AI4CCAM Ethical and Scientific Advisory Board who also mentioned AI4CCAM when dealing with different perspectives on AI and data in CCAM, from transversal AI developments that promise broad, cross-sector impact to tailored, application-driven solutions addressing specific mobility needs.
Bernhard Peischl (Innovation Manager, AVL List GmbH) was a further panellists.
A groundbreaking cluster of EU-funded projects has been formed to revolutionise road safety, automated mobility, and the interaction between drivers and vulnerable road users (VRUs). This collaborative effort brings together five ambitious initiatives – AI4CCAM, HEIDI, EVENTS, PHOEBE, and SOTERIA – to develop cutting-edge solutions addressing the growing complexity of urban transport systems, forming a cluster on “Road safety in complex urban environments”.
The cluster aims to promote a safe, inclusive, and sustainable mobility system that is resilient, trustworthy, and road user-centric. By uniting efforts across these projects, this initiative is set to transform European transport research and establish new standards for road safety and automated driving.
Cluster Overview
At the heart of the cluster’s mission is a shared vision for establishing the “safe system” approach. This shifts the focus from placing responsibility solely on road users to a holistic strategy where every stakeholder—from infrastructure designers to transport operators—plays a role in creating safer environments. The projects will work together to ensure that automated mobility technology is not only efficient but also transparent, inclusive, and adaptable to real-world road conditions.
The five projects are exploring how advanced technologies like artificial intelligence (AI), simulation environments, predictive analytics, and human-machine interfaces (HMIs) can enhance urban road safety for all road users, particularly vulnerable groups such as pedestrians, cyclists, and individuals with reduced mobility.
Projects Highlights
AI4CCAM leverages the potential of AI to create trustworthy and ethical models for predicting the behavior of vulnerable road users in urban environments. Its focus on user acceptance of automated vehicles and ethical dilemmas ensures the development of AI systems that people can trust.
EVENTS seeks to overcome the limitations of current Connected and Automated Vehicles (CAVs) by developing a robust and resilient perception and decision-making system that can manage unexpected “events” like adverse weather/light conditions, unstructured road environment, imperfect data, sensor/communication failures, etc., ensuring continuous safe operation in dynamic environments.
HEIDI is breaking new ground by designing a cooperative HMI that connects drivers and pedestrians in dangerous situations. With internal and external HMIs, HEIDI adapts in real time to the behaviors and needs of drivers and VRUs.
PHOEBE aims to support urban transport planning with an evidence-based framework for predictive road safety. This project offers a blueprint for cities to manage safety risks effectively, integrating human behavior modeling and transport system simulations to prevent accidents.
SOTERIA focuses on creating a data-driven safety intelligence framework that integrates electric micro-mobility services in urban environments. It emphasises inclusivity by fostering a co-creation process with local communities and vulnerable road users.
Collective Strengths
The cluster of projects is founded on several shared strengths:
Human-Centric and Inclusive: Prioritising the needs of all road users, including vulnerable populations such as children, elderly, and those with reduced mobility.
Ethical and Trustworthy AI: Developing AI models and decision-making systems that are transparent, reliable, and capable of handling complex ethical issues.
Advanced Simulation Technologies: Leveraging co-simulation environments, hybrid testing, and machine learning to safely evaluate and validate new technologies.
Scalability and Resilience: Ensuring that solutions are adaptable across various transport modes, from micro-mobility services to automated vehicles, and capable of handling unexpected events and system failures.
Impact and Future Directions
By integrating innovative technologies and road user-centered approaches, this EU-funded cluster of projects aims to deliver substantial advancements in road safety and automated mobility. The initiative will not only contribute to the development of safer transportation systems but also foster public trust and acceptance of emerging mobility solutions.
In particular, the exchange of knowledge and practices between projects regarding AI and simulation technologies could further amplify their positive impact, fostering innovation and improving outcomes across the board. Through close collaboration, these projects will offer new methodologies, standards, and tools to help urban planners, policymakers, and transport operators create safer and more efficient mobility networks across Europe.
On 16 October, a new online meeting was organised to talk about progresses and next steps the Use Case 3 (UC3) in AI4CCAM, mainly involving CNRS and TTS Italia in the discussion.
The purpose of UC3 is to assess the degree of Vulnerable Road Users’ (VRUs) acceptance of Connected and Automated Vehicles exploring their interaction dynamics in a number of increasingly complex urban settings, where users will be immersed through a purposely designed experiment. When interacting with CAVs, VRUs will be prompted to take decisions (e.g., street crossing decisions when a CAV is approaching), while semi-quantitative data, also including physiological indicators, will be collected to assess acceptance rates in diverse scenarios.
The recent meeting highlighted the continued progress on UC3 user acceptance, with a focus on refining urban traffic environments and addressing challenges related to CARLA tool constraints, an open-source autonomous driving simulator used in AI4CCAM to create scenarios and attempt experiences.
TTS Italia, leading this Use Case, is playing a pivotal role in defining the environment descriptions and scenario dynamics, focusing on how vulnerable road users (VRUs) and connected automated vehicles (CAVs) will interact during experiments. TTS is also working on the methodologies for measuring key metrics using advanced devices like EMG and GSR, ensuring precise data collection.
CNRS is actively working on creating optimized urban traffic environments that meet the project’s requirements, exploring alternative setups to ensure efficient communication metrics and enhancing the VR scenarios. Despite some challenges, such as device limitations and physiological measurement dynamics, CNRS is making headway in overcoming these obstacles.
IRTSX took part in the discussion too, contributing to enhance scenario dynamics description.
On the 10th of October, the CCAM (Connected, Cooperative, and Automated Mobility) Multicluster Meeting took place in Brussels, focusing on enhancing collaboration between large-scale demonstrations and various clusters within the CCAM Partnership.
AI4CCAM project, being a part of “Cluster 5: Key Enabling Technologies”, contributed to the discussion on how the project’s outcomes can support large-scale demonstrations across Europe.
The project contribution is centered on four major results expected from AI4CCAM:
- Methodology Input: Integration of EU Trustworthy AI guidelines into CCAM applications.
- Road Scenarios in MOSAR: Modelling of ethical issues within road scenarios.
- Participatory Space: Creation of a collaborative platform bringing together the general public and industry experts.
- Research on Scene Understanding: Contributions to pedestrian and vehicle trajectory prediction, including model definition and validation.
These project results have the potential to be scaled up and transformed into building blocks for large-scale demonstrations. Not only do they offer a qualitative understanding of scene dynamics and address ethical challenges in road scenarios, but they also provide an inclusive Participatory Space—designed to gather feedback from both users and citizens to enhance the public’s perception of CCAM technologies.
AI4CCAM presented the paper “Toward a Meet-in-the-Middle Methodology for Trustworthy AI for CCAM” during the last edition of the International Conference on Intelligent Traffic and Transportation (ICITT), held in Florence, Italy, 16-18 September 2024.
The International Conference on Intelligent Traffic and Transportation (ICITT) is a major event for academics, researchers, and industrialists who are engaged in Intelligent Traffic and Transportation research. Held annually since the late 2010, the conference is renowned as a friendly and inclusive platform that brings together a broad community of researchers who share a common goal: developing and managing engineering and technologies revolution of Transportation Systems, and operations which are key to sustaining the success of Intelligent Traffic and Transportation industries.
The publication presented by the AI4CCAM project, a collaboration between IRT SystemX, Barcelona Supercomputing Center (BSC), INLECOM, pushes the first iteration for a meet-in-the middle methodology for trustworthy Artificial Intelligence in Cooperative Connected and Automated Mobility (CCAM). In this result of AI4CCAM, IRT SystemX brings the scenario approach and a first pipeline of the methodology inspired in the Confiance.ai program, BSC brings the expertise on ethics and responsible AI and INLECOM leads the validation of use cases in the project that drive the modeled scenarios. The purpose is to leverage existing and proven practices as the integration of trustworthiness guidelines for AI is modeled and assessed.
The next CCAM Partnership Multi-cluster meeting is scheduled on 10 October 2024, in Brussels, and AI4CCAM, represented by the Arnaud Gotlieb, Simula Research Laboratory, the project coordinator, will be among the speakers fo the day.
The meeting will focus on the preparation of the future Work Programme 2026-27, especially the large-scale demonstration calls for projects.
AI4CCAM will be especially involved in the Breakout session on “What projects’ results can be integrated into large-scale demonstrations? What research needs can be postponed till the next Framework Programme?”, including CCAM Cluster 5 Group on Key Enabling Technologies, eager to learn more about the outcomes and results of the project, and understand how these could potentially be exploited within the future CCAM Large-scale demonstration projects.
On 25 and 26 September, AI4CCAM participated in the GSVF 2024 – Virtualization of Software-Defined Vehicles event, in Graz, Austria, organised by Virtual Vehicle, one of the project partners.
The vehicle records its surroundings with sensors and thermal imaging, “learns” to adapt its driving functions to this environment, uses intelligent navigation, and communicates with other AI-Defined Vehicles and intelligent infrastructure. What sounds futuristic is already a reality. AI algorithms enable real-time vehicle diagnostics and environmental perception. Essential to this is so-called Edge Computing – data processing within the vehicle itself. Because only in this way can cars react in real-time and much faster than a human ever could. Making vehicles smarter, safer, and more user-friendly is a core task at the Virtual Vehicle and was also a focus during their annual specialist conference, GSVF.
AI4CCAM was represented not only by Virtual Vehicle but also by a further project partner, CNRS, and presented both the work on the integration of virtual reality and CARLA, an open-source autonomous driving simulator used in AI4CCAM to create scenarios and attempt experiences.
This was a collaborative demonstration, that also served as an experiment, developed by the two teams together. Participants, after providing an informed consent, were presented with scenarios in virtual reality where they encountered various vehicles, in different whether and lighting conditions, and were required to judge when they would cross the road. The data collected in the demonstration/experiment will be used by CNRS to develop their behavioral model of pedestrian road-crossing behavior in the project.
AI4CCAM will take part in the event “Rethinking modal integration between Public Transport and shared, connected and autonomous mobility”, organised within the European project EMBRACER, in Vilnius, Lithuania, on 9 and 10 October.
EMBRACER is a region-driven project, funded within the Interreg Europe Program, where seven underserved regions have committed to integrate public transport with informal modes (cycling, ride-hailing, car/bike/scooter sharing, on-demand transport, autonomous shuttles) to enhance the interconnection with urban areas and achieve seamless intelligent climate-resilient regional and local intermodal mobility.
TTS Italia will be presenting the Use Case 3 of the AI4CCAM project, which explores user acceptance of automated vehicles equipped with Vulnerable Road User (VRU) prediction capabilities.
The Use Case aims to assess how VRUs interact with connected and automated vehicles (CAVs) before and after engaging in various traffic scenarios, such as T-junctions, crossroads, and roundabouts, within a VR environment. By measuring physiological indicators like GSR (Galvanic Skin Response) and EMG (Electromyography), AI4CCAM can evaluate changes in user acceptance, helping to enhance future interactions between CAVs and vulnerable road users.
Represented by Arnaud Gotlieb, Simula, project coordinator, AI4CCAM took part in MET’24 (Proceedings of the 9th ACM International Workshop on Metamorphic Testing), within the ISSTA/ECOOP event, held in Vienna, Austria, 16-20 September.
Simula had a double role, giving an invited Keynote speech on “AI-driven Metamorphic Testing for Autonomous Systems”; and also presenting the paper entitled “Evaluating Human Trajectory Prediction with Metamorphic Testing”.
The test oracle problem is one of the most fundamental and challenging problems in software testing and more broadly, software engineering. A growing body of research has examined the concept and approach of Metamorphic Testing (MT) and has justified that MT can effectively alleviate the oracle problem and detect real-life bugs in various application domains. Compared with most other testing methods where the correctness of each individual test output is checked, MT has a different perspective on testing: it focuses on the relationships among the inputs and outputs of multiple executions of the software under test.
MET: The International Workshop on Metamorphic Testing brought together researchers and practitioners in academia and industry to discuss research results, experiences, and insights into MT. The ultimate goal of MET is to provide a platform for the discussion of novel ideas, new perspectives, new applications, and the state of research, related to or inspired by MT.
Go to our Library and download the Keynote Speech and the paper!
