Researchers in the "SafeSign" project are currently addressing this question and more.
The "SafeSign" project is a one-year, exploratory project funded by the Austrian Research Promotion Agency as part of the Ideas Lab 4.0 program. Under the consortium leadership of RISC Software Ltd. and including the JKU and Austrian Road Operator ASFINAG, the issue is being researched and addressed to find better ways of using artificial intelligence to help self-driving vehicles read traffic signs.
Deep-learning methods will help evaluate the recognition system based on real traffic signs (with and without alterations as well as synthetically altered images) in order to subsequently use AI models to develop better interpretability. As ethical aspects play an increasingly important role in developing AI, Karin Bruckmüller and Stefan Schumann (both JKU) will explore and question ethical principles within the legally permitted parallels and apply them to the AI developments in the project. AIs able to best recognize traffic signs and subsequently respond as well as those created in an ethically appropriate way will increase the confidence level in new technologies for companies, individual users and the population as a whole.
The project findings for "SafeSign" will serve as an additional basis to create mobility for the future.