AITHENA project contributes to building Explainable AI (XAI) in Cooperative, Connected and Automated Mobility (CCAM). The project works on the development and testing frameworks, researching three main AI pillars: data, models, and testing.

The project will use a human-centric approach, where partners aim to derive trustworthy AI dimensions through user-identified needs in CCAM applications.

Demonstrators in the project will implement the AITHENA methodology through four critical Use Cases (UC):

Use case 1 – Perception: what does the AI perceive, and why.

Use case 2 – Situational awareness: what is the AI understanding about the current driving environment, including the driver state.

Use case 3 – Decision: why a certain decision is taken.

Use case 4 – Traffic Management: how transport-level applications interoperate with AI-enabled systems operating at vehicle-level.

AITHENA’s Use Cases have been shaped to address the general and specific goals defined for the project.

UC-1 Perception – Trustworthy perception systems for CCAM

Demonstrator: Reliable pedestrian detection in urban environments.

Trusted AI is needed to understand which objects are perceived, how different sensor information is used, how possible discrepancies are solved.

UC-2 Understanding – AI extended situation awareness and understanding

Demonstrator: Collision prediction with hybrid AI data fusion models.

AI needs extended situation awareness to function properly in CCAM solutions. To acquire meticulous and complete knowledge about its environment, it needs to understand map information, perception layers, communications and much more.

UC-3 Decision – Trustworthy and human understandable decision making

Demonstrator: Explainable and robust decision making (manoeuvre and trajectory).

Path planning and manoeuvre execution maximises safety, comfort, and eco-driving. The user understands why, when, and how a decision is taken.

UC-4 Traffic – AI-based traffic management

Demonstrator: AI models and traffic management.

AI models are analysed in macroscopic scenarios with interaction with other AI and non-AI systems. Trustworthiness is gained by benchmarking “good behaviour” of AI models in transport level contexts.