The content discusses the challenges of ensuring safety for future cyber-physical systems (CPS) that are characterized by high levels of automation, interconnectedness, and the use of artificial intelligence (AI). It introduces the concept of Dynamic Risk Management (DRM) as a vision for addressing these challenges.
The key aspects covered are:
Assuring Autonomy: Handling the complexity of highly automated systems by enabling them to dynamically assess their own safety-related capabilities (Dynamic Capability Assessment) and the risks in the current operational situation (Dynamic Risk Assessment).
Assuring Interconnectedness: Addressing the challenge of safety modularization in cooperative CPS, where safety-related properties of collaborating systems need to be dynamically assessed at runtime.
Assuring Machine Learning: Dealing with the uncertainties introduced by the use of AI-based perception components, by combining them with more traditional, assurable channels to ensure safety.
The article presents the conceptual architecture of DRM, which follows a MAPE-MART/K (Monitor-Analyze-Plan-Execute over a MART - Models at Runtime - and Knowledge base) approach. It then goes into more detail on the key building blocks of DRM:
The article concludes by highlighting the need for further research to fully realize the potential of DRM and enable the safe and performant operation of future cooperative automated CPS.
To Another Language
from source content
arxiv.org
Key Insights Distilled From
by Daniel Schne... at arxiv.org 05-07-2024
https://arxiv.org/pdf/2401.13539.pdfDeeper Inquiries