This week we introduce next AIthena consortium partner: Virtual Vehicle Research GmbH.
The Virtual Vehicle Research GmbH is Europe’s largest R&D centre for virtual vehicle technology whose partner network comprises nearly a hundred industrial partners and over forty scientific institutions, both national and international. Leveraging the collective expertise of our interdisciplinary teams and collaborative partnerships, our centre seeks to bridge the divide between the virtual and real world, linking numerical simulations and hardware testing. Connecting the two worlds allows for a powerful hardware-software system design and automation of testing and validation procedures.
What is the role Virtual Vehicle Research GmbH and the team in the project?
The Virtual Vehicle Research GmbH will assume the leadership role at the final stage of the ML (Machine Learning) models’ life cycle – the evaluation. Our team will provide expert guidance and actively contribute to the formulation of the testing and validation methodology. The validation methodology will serve to demonstrate the capabilities, identify the limitations, and assess the human-factor-specific KPIs of the developed CCAM systems.
However, our team’s involvement will extend across all life cycle stages, from the identification of stakeholders’ needs, expectations, and concerns, defining ML development frameworks, building safety-critical and trustworthy models, to preparing for physical testing. Ultimately, to ensure the success and broaden the project’s impact, we will engage in scientific and industrial dissemination through publications, conference presentations, and workshops.
What are you currently working on in the project?
Presently, our focus lies in the building and training of a trustworthy and robust XAI model, specifically developed for predicting the intended motion of traffic participants in complex urban environments. By combining physics-based and data-driven approaches, we aim to leverage the benefits of both in enhancing of the model’s transparency and extending of the prediction time horizons. As such, the developed model could be integrated into the robo-taxi, allowing its safe and predictable operation especially in scenarios involving VRUs (Vulnerable Road Users). Furthermore, our team has initiated preparations for our forthcoming leadership role in the model deployment and testing.
From your perspective, how do you see the contribution of the AIthena project to building trustworthy, explainable, and accountable AI-based CCAM?
Built upon the pillars of trustworthiness, AIthena project provides a foundation for all stages of the AI-based CCAM life cycle. The methodology adopted to enhance explainability, accountability, and fairness of the models within the use-cases can be extrapolated to include other CCAM applications. The example of the XAI model developed by our team can be applied to another application seeking to incorporate existing expert knowledge of the physics-based approaches and the complex problem-solving capabilities of the data-based approaches. The conclusion holds true for other use-case models, affirming the AIthena’s primary contribution as a blueprint for the future development of the CCAM systems.
You can find more about the Virtual Vehicle Research GmbH centre’s expertise and project involvement on About – Virtual Vehicle (v2c2.at).