Final Event
On 20 May 2026 the jbDATA Final Event took place in Schwalbach am Taunus, marking the successful completion of the project with around 80 participants from industry, research institutes, public authorities and the mobility community.
Final Event
We are pleased to invite you to the jbDATA Final Event on 20 May 2026 in Schwalbach am Taunus.
On 20 May 2026, just better DATA hosted its final event at Aumovio Germany GmbH in Schwalbach am Taunus, bringing together 80 participants for a day filled with insights into project results and collaborative workshops exploring future directions. Attendees experienced interactive demo sessions that brought tomorrow’s innovations to life by showcasing the Smart Data Loop hands-on, from edge-side operations within the vehicle and cloud‑based orchestration and processing through to continuous model improvement via retraining and deployment.
The event provided a valuable opportunity to reflect on the progress achieved and to celebrate the possibilities ahead together.
Over the past three years, just better DATA has pioneered the Smart Data Loop, a systematic and efficient approach to provide high quality data for the development, testing and validation of advanced automotive systems, including AI driven solutions for autonomous driving.
Join us for a day filled with insights into our project results and collaborative workshops exploring future directions. Experience interactive demo sessions that bring tomorrow’s innovations to life by showcasing the Smart Data Loop hands-on, from edge-side operations within the vehicle and cloud‑based orchestration and processing through to continuous model improvement via retraining and deployment.
Date: 20 May 2026
Location: Aumovio Germany GmbH, Sodener Str. 9, 65824 Schwalbach am Taunus
Format: In-person event
Cost: Free of charge
Let’s celebrate progress and possibilities together!
Agenda
| 8:15 | Registration & Welcome Coffee |
| 9:00 | Welcome by Consortia Lead Azarm Nowzad, Aumovio |
| 9:15 | Welcome by Funding Office Heiko Krause, TÜV Rheinland Group |
| 9:30 | Two Perspectives: Leading the way to End-2-End Data Selection Dr. Claudio Longo, Head of Innovation, Aumovio Prof. Dr. Fabian Flohr, Professor of Machine Learning, Hochschule München |
| 10:15 | jbDATA – The Project Subproject Leads |
| 11:30 | Lunch Break |
| 12:15 | Introduction: Data Loop and Demonstration Session |
| 12:30 | Demonstration Session: Walk through the Data Loop Experience interactive demo sessions that bring tomorrow’s innovations to life by showcasing the Smart Data Loop hands-on, from edge-side operations within the vehicle and cloud‑based orchestration and processing through to continuous model improvement via retraining and deployment. |
| 14:00 | Deep Dive & Workshop Overview |
| 14:15 | Deep Dive Sessions Deep Dive 1 - Smart Data Curation: Can we detect corner cases before they hit the road? Are you still training on yesterday’s failures or tomorrow’s unknowns? This session explores how corner cases can be detected, analyzed, and prioritized early by moving beyond closed‑loop validation toward open‑loop, data‑driven learning. Participants will gain insights into: - Detecting corner cases before real‑world deployment - Analyzing driving data through an ODD‑centric lens - Using VLMs and LLMs to structure, explain, and scale long‑tail learning Deep Dive 2 - Edge Software Stack and Cloud Integration How can we monitor, manage, and reconfigure a smart edge platform through cloud integration? This session provides a deep dive into the software stack on the smart edge platform and its seamless integration with the cloud. It explores how edge components generate, process, and expose metadata, and how these data streams are continuously transferred to the cloud for monitoring, analysis, and orchestration. Participants will gain insights into: - Core software components and services deployed on the edge platform - Continuous metadata streaming from edge to cloud - Cloud‑based monitoring, diagnostics, and lifecycle management of edge systems - Enabling adaptive reconfiguration and scalable operation across distributed edge devices - This workshop highlights how a tightly coupled edge–cloud architecture is essential for operating data‑driven, intelligent systems at scale. |
| 15:30 | Coffee Break |
| 16:00 | Workshop Sessions Workshop 1 - From Data to Dataset: How can we create high-quality, reusable datasets for AI? Using the jbDATA tools and dataset as examples, this workshop explores practical approaches to create high-quality datasets for AI development. It addresses challenges such as data selection, publication, and legal requirements, including GDPR. Participants will gain insights into lessons learned during the project and discover how the jbDATA dataset can support future research and industrial applications. Workshop 2 - Synthetic Data: A Strategic Framework for Modern AI This workshop invites participants to explore key topics shaping the future of synthetic data, including the development of robust synthetic data strategies—from selecting appropriate generation techniques to ensure reliable data validation. The session also addresses the extent to which synthetic LiDAR data can approximate real-world measurements and the methods used to validate it, as well as the capability of synthetic LiDAR simulations to accurately replicate real sensor behavior. Participants will gain insights into current challenges, lessons learned, and the potential of synthetic data for future research and industry use. |
| 17:15 | Outlook and Perspectives: From Data Sharing to Shared Intelligence - Unlocking Industrial Collaboration Lukas Elster, Persival Fabian Flohr, Hochschule München Jörg Reichardt, AUMOVIO Ole Salschneider, SafeAD Torsten Stiehm, DXC Luxoft |
| 17:45 | Acknowledgements & Farewell |
Final Event Results
- Research Vehicle “NOVA” – AI-based Data Aquisition Christina Sigl, Stefan Seidl, Florian Bauer, Alexander Hilgart, Bastian Schumacher, Robert Smirnow, Simon Steiger, Stefan Weber (b-plus technologies GmbH), Zahra Boroujeni, Torsten Stiehm, Luis-Maximiliano Cuevas Eiroa (Luxoft DXC Inc.)
- Mobile Calibration Rig and Multimodal Calibration Fabian Bally, Michael Schötz, Martina Schöll (Deggendorf Institute of Technology)
- jbX – THD Development Platform Martina Schöll, Michael Schötz, Fabian Bally, Patrick Kühnel, Thomas Limbrunner (Deggendorf Institute of Technology)
- From Crash Research to Training Data Michael Schötz (Deggendorf Institute of Technology)
- Experiments on Cross-Modal Person Re-ID Robert Aufschläger, Gautam Savaliya, Prof. Dr. Michael Heigl (Deggendorf Institute of Technology)
- PriMod4AI: Lifecycle-Aware Privacy Threat Modeling for AI Systems Using LLMs Gautam Savaliya, Robert Aufschläger, Prof. Dr. Michael Heigl (Deggendorf Institute of Technology)
- Object-Level OoD Detection using Semantic Segmentation and Anomaly Scoring Alaeddin Abdellaoui, Azarm Nowzad (Aumovio SE)
- Scene-Aware Detection of Out-of-Distribution (OoD) Scenarios for Smart Data Curation Alaeddin Abdellaoui, Azarm Nowzad (Aumovio SE)
- Understanding Autonomous Driving Datasets by Describing Differences between Image Subsets in Natural Language Julian Truetsch (FZI)
- Feedback-Controlled Data Collection from Data Streams Philipp Reis, Philipp Rigoll (FZI Forschungszentrum Informatik)
- A Domain-Specific Language for LLM-Driven Trigger Generation in Multimodal Data Collection Philipp Reis, Philipp Rigoll (FZI Forschungszentrum Informatik)
- Context Aware Body Anonymization Kevin Rösch (FZI Forschungszentrum Informatik)
- Smart Data Loop in Autonomous Driving: Defining the Essential Requirements Felix Embacher, Julian Wiederer (Mercedes-Benz AG)
- Closing the Loop: Combined Edge-Cloud Approach to Minimize Imbalance in Datasets
Maximiliano Cuevas, Nenad Ocelic, Torsten Stiehm (DXC-Luxoft) - Agentic Scenario and Synthetic Data Generation Based on Requirements Zahra Boroujeni, Torsten Stiehm, Vinay Pyati (DXC-Luxoft)
- Criticality Detection for Selective Data Capture Dinish K Karunakaran, Shady Hassan, Max Nestoriuc (AVL Deutschland GmbH)
- Anomaly Detection for Selective Data Capture Using LSTM Autoencoders Dinish K Karunakaran, Shady Hassan, Max Nestoriuc (AVL Deutschland GmbH)
- Pseudo Label Generation and Data Engine Wencan Huang (Aumovio SE)
- Data Mining Strategy Yue, Yao (Aumovio SE), Michael, Schötz (Deggendorf Institute of Technology)
Two Perspectives: Leading the way to End-2-End Data Selection
Beyond justbetterDATA – From Data Strategy to Autonomous Reliability at Scale
Dr. Claudio Longo (Aumovio)
Beyond the Ghost in the Log – Building Reactive Closed-Loop Simulations
Prof. Dr. Fabian Flohr (Hochschule München)
Deep Dive 1
Philipp Reis (FZI Forschungszentrum Informatik), Nenad Ocelić (DXC Luxoft), Erwin Kraft (Aumovio)
Deep Dive 2
Michael Schötz (THD), Fabian Bally (THD), Robert Smirnow (b-plus technologies GmbH)
Workshop 1
From Data to Dataset: How can we create high-quality, reusable datasets for AI?
Hendrik Königshof (FZI), Philip Schörner (FZI), Peter Szabados (Luxoft), Yue Yao (Aumovio), Michael Schötz (THD)
Workshop 2
Synthetic Data: A Strategic Framework for Modern AI
Sandeep Kumar (Valeo), Nicolas Scheller (Valeo), Erwin Kraft (Aumovio), Torsten Stiehm (Luxoft), Jithin Sid (AVL S&F)
Videos will follow soon