Workshop on Data sharing for Automated Driving

On February 25^th, ARCADE organised an online workshop on the topic of data sharing in the CCAM area. ARCADE is a H2020 Coordination and Support Action. Data sharing is one of its topics. In the CAD knowledge base, ARCADE collected material for data sharing.

For cost-effective evaluation of CCAM and for comparability of automated driving research studies, data sharing is essential. If not, costly parallel work on data collection remains common practice. For training of AI, large data sets are needed as well. Many hurdles for data sharing exist, starting from data sensitivity, security, data formats and descriptions, to privacy (GDPR). Today’s working-from-home practice has quickly enabled some remote access facilities.

The importance of data sharing was also identified in the SRIA 1.0 (Cluster 5: System architecture for data sharing; Cluster 7: Common test data sharing framework).

In this workshop, important bottlenecks were identified and promising approaches for data sharing were presented. The morning session was a broader webinar presenting today’s status, experiences and challenges. The afternoon expert workshop featured three break-out sessions on defining boundary conditions and next steps following promising approaches.

The break-out sessions were:

1. Principles for industry data sharing
   What principles can we agree upon for enabling data sharing in industrial and research projects? Under what conditions is sharing possible? After an introduction of an international multi-partner project, promising approaches were presented.
2. In-vehicle data selection and analysis
   Data is generated in vehicles for research and development. This could lead to difficulties in managing large amounts of data, governing and protecting IP rights, and privacy issues. What if the data could be processed in-vehicle and only a reduced dataset is sent back-office, after data is pseudonized and reduced in size? This workshop discussed how we can reach this state and what are the steps needed.
3. GDPR in practice
   What practical privacy challenges do researchers and developers face in order to collaborate and share personal data? Examples of the type of problems which may be encountered were discussed, examining how (and if) they may be overcome.

For more information, please contact Sytze Kalisvaart TNO or Erik Svanberg SAFER.