The mission of openGENESIS is to provide knowledge, methods and tools for the assessment of artificial intelligence (AI) used for autonomous driving. Learning algorithms should be proved to be safe and roadworthy before deployed to public road. From today’s point of view, we are not able to completely understand and explain the detailed functionality of learned algorithms due to the high complexity. openGENESIS will supply public and regulation authorities with necessary approaches to deal with the challenges of approval and certification of AI. Providing a “TÜV for AI”. openGENESIS provides a collaboration platform for leveraging knowledge among members. The members cooperate with each other efficiently and provide the results in an open access domain. A strong global exchange between industry, research and regulators will be established. Arrive at a common understanding of quality aspects for artificial intelligence.
openGENESIS drives and supports the investigations of a verifiable, understandable and not least certifiable AI. This includes activities of adopting functional safety approaches to symbolic and subsymbolic artificial intelligence as well as the development of new methods and metrics for a formal verification of neuronal networks. Approaches regarding a better understandability as well as reproducibility of the decisions of AI will take place. An open data set will be established and research should be enabled with direct access to practice and safety relevant challenges. Finally, also social concerns should be considered in aspects for a certification of learned algorithms.
- DFKI (German Research Center for Artificial Intelligence)
- TÜV SÜD
Current Status (13.09.2018)
The openGENESIS Charta is drafted and currently in internal review process. It will be published for joint review in septembre 2018.