AI4CCAM – Trustworthy AI for Connected, Cooperative and Automated Mobility.
To develop an open environment for integrating trustworthy-by-design AI models of vulnerable road user behavior anticipation in urban traffic conditions.
The AI4CCAM project aims at addressing trustworthiness of artificial intelligence in the context of CCAM (Connected, Collaborative and Automated Mobility). The project focusses on urban scenarios and interactions with Vulnerable Road Users as ethical, social and cultural implications of trustworthiness of AI are assessed.
AI4CCAM relies on a methodology that tackles the following challenges:
- Development of a digital framework for CCAM scenario management. Namely scenarios for interactions between VRUs and automated vehicles: their generation, modelling and management. A special focus is made on interaction with multiple VRUs and related ethical dilemmas as well as acceptance. Systematic synthetic AI-based scenarios covering these angles is insufficient and should be addressed in the project.
- Development of improved VRU behavior anticipation models based on visual gaze estimates in AI models.
- Design of trustworthy-by-design AI AI-based decision-making models. In this scope, explainability and robustness in design, training and deployment should be considered.
- User acceptance challenges through identification of user-driven and data-driven requirements on diversity and fairness in CCAM applications.
IRT SystemX will contribute to the development of a methodology for trustworthy AI for CCAM using the MOSAR approach for scenario management, as well as to the development of trustworthy by design AI-based predictive behaviour models for VRUs, while ensuring the overall Technical Coordination for the project.
AI4CCAM project is funded by the European Union under the Horizon Europe Research and Innovation Program (Grant Agreement n°101076911).
- Methodology for trustworthy AI in CCAM
- AI4CCAM Trustworthy AI documentation framework
- Interoperable data-driven AI4CCAM digital framework
- AI4CCAM models for VRUs movement anticipation and ego car trajectory prediction
- Explainable AI tools for VRU detection
- AI4CCAM validation handbook
- CAV-VRU interaction demonstrator
- Participatory AI4CCAM Space
- Data-driven simulation scenarios
|Data Science and IA|
- CCAM-related markets : OEMs ; regulators ; equipment and software provider…