Machine learning introduces new safety challenges for autonomous vehicles, especially when traditional safety standards fall short. This presentation shows how SOTIF can be used to identify and mitigate risks tied to functional insufficiencies and unknown scenarios in ML-based systems. It’ll cover how to build safety arguments for ML, address uncertainty, and ensure robust performance across diverse environments.
In this session, you will learn more about:
- How to apply SOTIF (ISO/PAS 21448) principles to machine learning-based perception and control systems
- Methods for constructing structured safety arguments that address the limitations and uncertainties of ML models
- Best practices for identifying unknown unsafe scenarios and improving ML model robustness across diverse operational domains
&nsbp;
