On May 20, 2026, the final event for the just better DATA project took place at AUMOVIO in Schwalbach am Taunus. Around 80 participants gathered to review the results of the past three years, discuss them together, and look ahead to the future.

A Success Story: jbDATA and the Smart Data Loop

Over the course of the three-year research project, jbDATA developed the Smart Data Loop—a systematic and efficient approach to providing high-quality data for the development, testing, and validation of modern perception systems. The focus was primarily on AI-based applications and solutions for automated driving.

The Smart Data Loop connects all relevant processes along the data-driven development chain:

• Data collection and processing in the vehicle (edge);
• Cloud-based processing and analysis of the data;
• Supplementing underrepresented data using synthetic data generation;
• Continuous improvement of models through retraining and deployment in a closed loop.

This closed loop enables data to be collected more efficiently, thereby continuously optimizing AI models.

Focus on Interactive Formats

A highlight of the event was the interactive sessions, which provided participants with opportunities for exchange, discussion, and practical insights.

The workshop From Data to Dataset explored how to create high-quality, reusable datasets for AI applications. At the same time, there was a workshop on Synthetic Data. There, the focus was on a strategic framework for the use of synthetic data in modern AI systems.

Live Demonstrations: Innovation you can Experience Firsthand

The demo sessions vividly illustrated how the Smart Data Loop can work in practice – from hardware integration in the vehicle to cloud-based processing. This made complex interrelationships tangible and highlighted concrete application possibilities. Participants were able to immerse themselves on-site in vehicle technology, including sensor systems and an edge platform. Based on the Data Loop concept, they could explore a calibration setup, discuss trigger-based data collection, and experience anomaly detection live. In addition, data processing in the cloud and the anonymization of live data were demonstrated. The session was accompanied by a poster exhibition, where the project’s experts showcased additional findings and were able to answer questions from interested guests in detail.

Technical Presentations and Insights into Key Topics

Key project findings were presented in numerous presentations. Additionally participants could hear external keynote presentations on the topic of end-to-end data selection by Dr. Claudio Longo (AUMOVIO) and Prof. Dr. Fabian Flohr (Hochschule München).

In Deep Dives, participants gained technical insights into the topics of Smart Sata Curation as well as Edge Software Stack and Cloud Integration.

The program was diverse in scope. It comprehensively reflected the wide range of the project’s content – from the hardware used at the edge, through AI-based data selection and the targeted supplementation with synthetically generated data, all the way to the closed data loop. At the same time, it illustrated the close integration of research and industrial application.

The Conclusion: From Data Sharing to Shared Intelligence

A highlight of the event was the panel discussion at the end of the day. With experts from the startup scene, industry, and research, the discussion focused on other projects and players in the data ecosystem, but above all, it explored the potential of the jbDATA results and the dataset to be published. Furthermore, the discussion, moderated by consortium lead Azarm Nowzad (AUMOVIO), highlighted how collaboration and data utilization can create new potential for value creation.

Collaboration as the Key to Success

A key factor in the success of jbDATA was the close collaboration within a strong network of partners from industry and research. The intensive exchange between the partners and the event participants highlighted how important collaborative approaches are for the further development of data-driven innovations and how joint research can lead to concrete solutions for forward-looking technologies. The approaches developed – most notably the Smart Data Loop – provide a strong foundation for the next generation of end-to-end data-driven vehicle systems.

A key outcome of the project is the planned release of the jbDATA dataset. As a high-quality, practical dataset, it is intended to provide long-term support for the further development of data-driven applications, particularly in the fields of AI and automated driving and will be made available to the community. In addition, a comprehensive white paper is scheduled for publication.

Partners of jbDATA: AUMOVIO, AVL Deutschland GmbH, AVL Software and Functions GmbH, b-plus technologies GmbH, DXC Luxoft, FZI Forschungszentrum Informatik, Mercedes-Benz AG, Technische Hochschule Deggendorf und Valeo Schalter und Sensoren

Pictures: Max-Moments